CN1290008C - Virtual networking system and method in processing system - Google Patents

Virtual networking system and method in processing system Download PDF

Info

Publication number
CN1290008C
CN1290008C CN 02811128 CN02811128A CN1290008C CN 1290008 C CN1290008 C CN 1290008C CN 02811128 CN02811128 CN 02811128 CN 02811128 A CN02811128 A CN 02811128A CN 1290008 C CN1290008 C CN 1290008C
Authority
CN
China
Prior art keywords
processor
logic
virtual
network
node
Prior art date
Application number
CN 02811128
Other languages
Chinese (zh)
Other versions
CN1520550A (en
Inventor
P·舒尔特
S·金恩
P·曼卡
P·柯蒂斯
E·米尔尼
M·史密斯
A·格林斯潘
E·达菲
V·布劳内尔
B·斯普拉齐曼
D·布斯比
Original Assignee
伊金耐勒股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority to US28529601P priority Critical
Priority to US10/037,191 priority patent/US20030130832A1/en
Priority to US10/038,353 priority patent/US7231430B2/en
Priority to US10/038,355 priority patent/US6971044B2/en
Priority to US10/038,354 priority patent/US7174390B2/en
Application filed by 伊金耐勒股份有限公司 filed Critical 伊金耐勒股份有限公司
Publication of CN1520550A publication Critical patent/CN1520550A/en
Application granted granted Critical
Publication of CN1290008C publication Critical patent/CN1290008C/en

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00Data switching networks
    • H04L12/28Data switching networks characterised by path configuration, e.g. local area networks [LAN], wide area networks [WAN]
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L29/00Arrangements, apparatus, circuits or systems, not covered by a single one of groups H04L1/00 - H04L27/00
    • H04L29/12Arrangements, apparatus, circuits or systems, not covered by a single one of groups H04L1/00 - H04L27/00 characterised by the data terminal
    • H04L29/12009Arrangements for addressing and naming in data networks
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L29/00Arrangements, apparatus, circuits or systems, not covered by a single one of groups H04L1/00 - H04L27/00
    • H04L29/12Arrangements, apparatus, circuits or systems, not covered by a single one of groups H04L1/00 - H04L27/00 characterised by the data terminal
    • H04L29/12009Arrangements for addressing and naming in data networks
    • H04L29/12018Mapping of addresses of different types; address resolution
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/35Application specific switches
    • H04L49/351LAN switches, e.g. ethernet switches
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L61/00Network arrangements or network protocols for addressing or naming
    • H04L61/10Mapping of addresses of different types; Address resolution
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network-specific arrangements or communication protocols supporting networked applications
    • H04L67/10Network-specific arrangements or communication protocols supporting networked applications in which an application is distributed across nodes in the network
    • H04L67/1002Network-specific arrangements or communication protocols supporting networked applications in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers, e.g. load balancing
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network-specific arrangements or communication protocols supporting networked applications
    • H04L67/10Network-specific arrangements or communication protocols supporting networked applications in which an application is distributed across nodes in the network
    • H04L67/1002Network-specific arrangements or communication protocols supporting networked applications in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers, e.g. load balancing
    • H04L67/1004Server selection in load balancing
    • H04L67/1017Server selection in load balancing based on a round robin mechanism
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network-specific arrangements or communication protocols supporting networked applications
    • H04L67/10Network-specific arrangements or communication protocols supporting networked applications in which an application is distributed across nodes in the network
    • H04L67/1097Network-specific arrangements or communication protocols supporting networked applications in which an application is distributed across nodes in the network for distributed storage of data in a network, e.g. network file system [NFS], transport mechanisms for storage area networks [SAN] or network attached storage [NAS]
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network-specific arrangements or communication protocols supporting networked applications
    • H04L67/38Protocols for telewriting; Protocols for networked simulations, virtual reality or games
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L69/00Application independent communication protocol aspects or techniques in packet data networks
    • H04L69/30Definitions, standards or architectural aspects of layered protocol stacks
    • H04L69/32High level architectural aspects of 7-layer open systems interconnection [OSI] type protocol stacks
    • H04L69/322Aspects of intra-layer communication protocols among peer entities or protocol data unit [PDU] definitions
    • H04L69/329Aspects of intra-layer communication protocols among peer entities or protocol data unit [PDU] definitions in the application layer, i.e. layer seven
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L69/00Application independent communication protocol aspects or techniques in packet data networks
    • H04L69/40Techniques for recovering from a failure of a protocol instance or entity, e.g. failover routines, service redundancy protocols, protocol state redundancy or protocol service redirection in case of a failure or disaster recovery
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/35Application specific switches
    • H04L49/354Support for virtual LAN, VLAN tagging or multiple registration, e.g. according to IEEE 802.1q
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/55Error prevention, detection or correction
    • H04L49/557Error correction, e.g. fault recovery or fault tolerance
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network-specific arrangements or communication protocols supporting networked applications
    • H04L67/10Network-specific arrangements or communication protocols supporting networked applications in which an application is distributed across nodes in the network
    • H04L67/1002Network-specific arrangements or communication protocols supporting networked applications in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers, e.g. load balancing
    • H04L67/1034Reaction to server failures by a load balancer

Abstract

一种用于虚拟连网(100)的系统和方法。 For a virtual networking (100) for a system and method. 在底层的点对点网格上提供交换以太网局域网的语义。 Switched Ethernet LANs provide semantic point on the bottom of the grid. 计算机处理器节点能通过交换结构的虚拟接口直接通讯,或者它们能通过以太网交换模拟器(115a,115b)通讯。 The computer processor nodes via virtual direct communication switch fabric interfaces, Ethernet or they can exchange simulator (115a, 115b) communications. 使用ARP(地址解析协议)逻辑(135)将IP地址与虚拟接口关联,同时允许计算机处理器回答有虚拟MAC地址的ARP请求。 Using ARP (Address Resolution Protocol) logic (135) the IP address associated with the virtual interface, while allowing the computer processor to answer the ARP request with a virtual MAC address.

Description

在处理系统中虚拟连网的系统及方法 System and method for processing system in a virtual networking

背景1.发明的领域本发明涉及用于企业及应用服务提供者的计算系统,更特别地,涉及具有虚拟化的通讯网络的处理系统。 BACKGROUND 1. Field of Invention The present invention relates to the application service provider enterprise and computing systems, and more particularly, relates to a processing system having a communication network virtualization.

2.相关技术的讨论在当前企业的计算和应用服务提供者的环境中,来自多个信息技术(IT)功能领域(电的,连网中的等)的个人必须参与部署处理及连网资源。 2. Discussion of Related technology in the current enterprise computing and application service provider environment, from multiple functional areas of information technology (IT) (electricity, networking, etc. in) the individual must be involved in the deployment processing and networking resources . 因此,由于在协调多个部门的活动方面的调度及其他困难,部署新的计算机服务器可能花费数周或数月时间。 Therefore, since the scheduling and other difficulties in coordinating the activities of various departments of the aspects of the deployment of new computer server may take weeks or months. 此冗长的手动过程增加了人力及设备开销,并延迟了应用的启用。 This lengthy manual process increases the cost of manpower and equipment, and delayed-enabled applications.

此外,由于难以预计应用将需要多少处理能力,经理通常过高提供计算能力的量。 In addition, how much processing power is difficult to predict because the applications will require, managers are often too high to provide the amount of computing power. 其结果是,数据中心计算资源常常未使用或使用不足。 As a result, data center computing resources are often unused or underutilized.

若最终需要的处理能力大于原先提供的,再次需要各种IT功能来协调活动以配置更多或更好的服务器,将它们连接到通讯和存储网络等。 If ultimately required processing power is greater than previously provided, once again the need to coordinate the activities of various IT functions more or better to configure the server, connect them to the communications and storage networking. 当系统越来越大时,此任务越来越困难。 When the system is growing, this task increasingly difficult.

部署也是成问题的。 Deployment is also problematic. 例如当部署24个常规的服务器,可能需要超过100个独立的连接来配置整个系统。 For example, when a conventional deployment server 24, may require more than one hundred separate connections to the entire system configuration. 管理这些电缆也是面临的挑战,每个电缆意味着一个故障点。 These cables are also management challenges faced by each cable means that a point of failure. 通过增加冗余来减轻故障的危险的企图使连线倍增,加重了问题且增加了复杂性和开销。 To mitigate the risk of failure of the attempt to make the connection by adding redundant doubled, adding to the problem and increases the complexity and overhead.

用当今的技术提供高可用性是困难和高花费的建议,通常对每个主服务器必须部署故障在线恢复服务器。 Provide high availability with today's technology is recommended difficulties and high cost, often must be deployed for each primary server failure recovery server online. 此外,常常需要复杂的管理软件和专业性服务。 In addition, often require complex management software and professional services.

通常,在传统的服务器上不可能调节处理能力或升级CPU。 Typically, the server is not possible in traditional conditioning capacity or upgraded CPU. 相反,缩放处理能力和/或迁移到厂商的新一代体系中常常需要“铲车升级”,意味着需要更多的硬/软件系统,需要新的连接等。 Instead, the scaling processing power and / or migration to next-generation system vendors often require "forklift upgrade" means more hardware / software systems, the need for new connections.

因此,需要为企业和ASP计算提供平台的系统和方法,致力于解决上述缺点。 Therefore, it is necessary to provide a system and method platform for enterprise computing and ASP, to solve the above drawbacks.

概述本发明的特征在于用于计算处理的平台与方法,在其中可以配置及部署虚拟处理区域网络。 Summary of the Invention wherein the platform and the method for calculation processing in which the processing may be configured and deployed virtual LAN.

按本发明的一个方面,提供用于模拟交换以太局域网的方法和系统。 According to one aspect of the present invention, there is provided a method and system for simulating switched Ethernet LANs. 提供多个计算机处理器和交换结构结构和到处理器的点对点的链路。 Providing a plurality of computer processors and the switch fabric and configuration of point to point link to the processor. 虚拟接口逻辑在交换结构结构和点对点链路上建立虚拟接口。 Virtual interface logic to establish a virtual interface on the switch fabric configuration and point link. 每个虚拟接口定义从一个计算机处理器到另一个计算机处理器经过交换结构结构的软件通讯路径。 Each virtual interface definition from a computer processor to another processor computer software structure as the communication path through the exchange. 以太网驱动器模拟逻辑至少在两个计算机处理器上执行,而交换模拟逻辑至少在一个计算机处理器上执行。 Ethernet driver on analog logic to perform at least two computer processor, performs analog logic and switching on at least one computer processor. 交换模拟逻辑在交换模拟逻辑和每个具有在其上执行的以太网驱动器模拟逻辑的计算机处理器之间建立虚拟接口,以允许它们之间的软件通讯。 Analog logic switching exchange and each analog logic between the Ethernet driver having a logic simulation executing thereon computer processor to establish a virtual interface software to allow communication therebetween. 它也从到具有在其上执行的以太网驱动器模拟逻辑的计算机处理器的一个虚拟接口接收报文,并发送该报文到另一个具有在其上执行的以太网驱动器模拟逻辑的计算机处理器,以响应与该报文相关的寻址信息。 It is also to have a from its analog logic Ethernet driver on a computer processor executing a virtual interface to receive packets, and send the packet to another Ethernet driver having a logic simulation on a computer processor executing in response to the message associated with the addressing information. 它也在每个具有在其上执行的以太网驱动器模拟逻辑的计算机处理器和每个另外的具有在其上执行的以太网驱动器模拟逻辑的计算机处理器之间建立虚拟接口。 It is also established between the virtual interfaces each having its Ethernet driver logic simulation on a computer processor executing a computer processor and each further having an Ethernet driver logic simulation is performed thereon. 如果虚拟接口运行良好,以太网驱动器模拟逻辑与模拟的以太网中另外的计算机处理器通过在它们之间定义软件通讯路径的虚拟接口单播通讯,若虚拟接口运行并不良好,则通过交换逻辑单播通讯。 Works well if the virtual interface, an Ethernet driver Ethernet analog logic and analog computer processor additionally by a virtual interface software defined unicast communication paths of communication between them, if the virtual interface is not operating well, through swap logic unicast communications.

按本发明的另一方面,提供实现地址解析协议(ARP)的方法和系统。 According to another aspect of the present invention, there is provided a method and system implementation Address Resolution Protocol (ARP) is. 计算平台具有多个由底层物理网络连接的处理器。 A computing platform having a plurality of processors are connected by the underlying physical network. 在处理器中的一个上执行的逻辑定义要在计算平台上模拟的以太网的拓扑结构。 Defined logic executed on a processor on a computing platform to be simulated topology Ethernet. 拓扑结构包括处理器节点和交换节点。 Topology including processor nodes and switching nodes. 在处理器中的一个上执行的逻辑从多个处理器中指定一组处理器作为处理器节点的处理器。 Logic executed on the processor to specify a group of processors as a processor from the plurality of processor node processors. 在处理器中的一个上执行的逻辑对模拟的以太网的每个处理器节点指定虚拟MAC地址。 Logic executed on the processor to specify a virtual MAC address for each processor node of the Ethernet analog. 在处理器中的一个上执行的逻辑在底层物理网络上分配虚拟接口,以提供从每个处理器节点的每个其他处理器节点的直接软件通讯。 Logic executed on a processor assigned virtual interface on the underlying physical network software to provide direct communication from each of the other processor nodes each processor node. 每个虚拟接口具有对应的标识符。 Each virtual interface has a corresponding identifier. 每个处理器节点具有ARP请求逻辑,将ARP请求传送到交换节点,其中ARP请求包括IP地址。 Each logical processor node having the ARP request, the ARP request is transmitted to the switching node, wherein the ARP request including the IP address. 交换节点包括ARP请求广播逻辑,将ARP请求传送到模拟的以太网中所有其它处理器节点。 Broadcasting an ARP request including switching node logic, an ARP request is transmitted to all other processor nodes in Ethernet analog. 每个处理器节点具有ARP回答逻辑,以确定它是否为与在ARP请求中的IP地址有关的处理器节点,若是,向交换节点发出ARP回答,其中ARP回答包含与IP地址相关的处理器节点的虚拟MAC地址。 Each ARP reply processor node having logic to determine whether it is a processor node an IP address in the ARP request related, if so, an ARP reply sent to the switching node, wherein the ARP reply comprises a processor node an IP address associated with the the virtual MAC address. 交换节点包括ARP回答逻辑,以便接收ARP回答并修改ARP回答,使其包括对该ARP请求节点的虚拟接口标识符。 ARP reply switching node comprises logic to receive and modify ARP reply ARP reply, to include an identifier of the virtual interface ARP requesting node.

按本发明的另一方面,提供计算机处理的平台和方法,以支持处理器的故障在线恢复。 And internet by another aspect, the present invention provides a computer processing, the processor to support fault recovery line. 多个计算机处理器被连接到内部通讯网络。 More computer processors are connected to the internal communications network. 在内部网络上定义并建立虚拟的局域通讯网络。 Define and create a virtual local area network communication on the internal network. 在虚拟的局域通讯网络中的每个计算机处理器具有对应的虚拟MAC地址,而虚拟局域网提供在一个计算机处理器组中的通讯,但不包括不在所确定的组中的多个处理器的通讯。 Each processor in the virtual local area computer communications network has a corresponding virtual MAC address, and a virtual local area network provides communications computer processor group, but does not include the group is not determined in a plurality of processors communication. 用对存储网络的地址空间所定义的对应关系定义并建立虚拟存储空间。 It defines the relation with the address space of the storage network, as defined and established virtual storage space. 响应计算机处理器的故障,从多个处理器中分配一个计算机处理器以代替故障的处理器。 In response to failure of a computer processor, a computer processor to assign a plurality of processors in place of the failed processor. 故障处理器的MAC地址被赋给替代故障处理器的处理器。 MAC address of the failing processor is assigned to replace a failed processor to a processor. 虚拟存储空间和故障处理器的确定的对应关系被赋给替代故障处理器的处理器。 Determining correspondence between the virtual memory space and processor failing processor is assigned to replace the failed processor. 虚拟局域网被重新建立,以包括替代故障处理器的处理器并排除故障的处理器。 Virtual LAN is re-established, in order to replace the failed processor including a processor and troubleshooting processor.

按本发明的另一方面,提供由IP地址编址的服务的系统和方法。 According to another aspect of the present invention, there is provided a system and method for addressing an IP address of the service. 至少有两个计算机处理器,每个包括提供服务的逻辑。 A computer with at least two processors, each comprising logic to provide services. 群集逻辑接收对服务的请求报文。 Cluster logic receives a request message for a service. 该报文具有IP地址。 The packet has an IP address. 群集逻辑将请求分配给具有提供该服务的逻辑的至少两个计算机处理器的一个。 A computer processor, the at least two clusters allocated to a requesting logic providing the service logic is.

按本发明的另一方面,计算机处理平台包括多个连接到内部通讯网络的计算机处理器。 According to another aspect, a computer processing platform according to the present invention comprises a plurality of internal communications network connected to a computer processor. 至少一个控制节点与外部通讯网络以及具有外部存储地址空间的外部存储网络通讯。 The at least one control node and an external communication network and a communication network external memory of the external memory address space. 至少一个控制节点连接到内部网络,从而与多个计算机处理器通讯。 At least one control node is connected to the internal network, whereby communication with a plurality of computer processors. 配置逻辑从多个处理器中定义并建立具有对应的计算机处理器组的虚拟处理区域网络;定义并建立在该计算机处理器组中提供通讯但排除未定义在该组内多个处理器的通讯的虚拟局域通讯网,以及与存储网络的地址空间有已定义的对应关系的虚拟存储空间。 Configuration logic to define and establish a virtual network having a processing region corresponding to the group of computer processors from the plurality of processors; defining and establishing communication in the computer processor to provide the group is not defined but excluding the communication within the group of the plurality of processors virtual local area communications network, as well as address space storage network virtual memory space corresponding relationship has been defined.

附图简述图1是示出本发明的实施例的系统的示意图;图2A-C是示出按本发明的实施例建立的通讯链路的示意图;图3A-B是示出本发明的某些实施例的网络软件结构的示意图;图4A-C是示出按本发明的某些实施例的驱动器逻辑的流程图;图5示出按本发明的某些实施例的服务群集; BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic diagram illustrating an embodiment of the system of the present invention; Figures 2A-C are schematic diagrams illustrating a communication link according to embodiments of the present invention established; Figures 3A-B are diagrams illustrating the present invention some network schematic software configuration of the embodiment; Figures 4A-C is a flowchart illustrating driver logic according to some embodiments of the present invention; FIG. 5 shows a clustered service according to some embodiments of the present invention;

图6示出本发明的某些实施例的存储软件的结构;图7示出本发明的某些实施例的处理器端存储逻辑;图8示出本发明的某些实施例的存储地址映射逻辑;和图9示出本发明的某些实施例的群集管理逻辑。 FIG. 6 shows the structure of some of storing software embodiment of the present invention; FIG. 7 shows the logical storage processor end some embodiments of the present invention; FIG. 8 shows a memory address some embodiments of the present invention, a mapping logic; and FIG. 9 shows a certain cluster management logic of the embodiment of the present invention.

详细描述本发明的首选实施例提供一处理平台,从中虚拟系统能通过配置命令被部署。 The invention is described in detail the preferred embodiment provides a processing platform, virtual system which can be deployed configuration commands. 该平台提供大的处理器池,从中可通过软件命令选择并构造子集以形成虚拟化的计算机网络(“处理区域网”或“处理器群集”),它们能部署成服务于一组给定的应用或客户。 The platform provides a large pool of processors, which may be selected by a software command and configured to form a subset of the virtual computer network ( "processing area network" or "cluster processors"), which can be deployed to serve in a given set application or client. 虚拟化的处理区域网(PAN)然后能用于执行针对客户的应用,如基于Web的服务器的应用。 Virtualized processing area network (PAN) can then be used to perform for the customer's applications, such as Web-based application server. 虚拟化能包括局域网(LAN)的虚拟化或I/O存储的虚拟化。 Virtualization can include a local area network (LAN) or a virtual I / O storage virtualization. 借助提供那样的平台,能很快并容易地通过软件,经过如来自管理员的配置命令部署处理资源,向每个服务器等等提供处理能力,而不必实际地提供服务器,电缆网络或存储连接。 With that kind of platform to provide, quickly and easily through software, such as through configuration commands from the administrator to deploy processing resources to provide processing power to each server, etc., without actually providing server, storage or network cable connection.

平台及其行为的概述如图1所示,首选硬件平台100包括经高速互连110a,b连接到交换结构结构115a,b的一组处理节点105a-n。 Overview of the platform and its behavior shown in Figure 1, the preferred platform 100 includes hardware via a high speed interconnect 110a, b is connected to the switch fabric structures 115a, b, a set of processing nodes 105a-n. 交换结构115a,b也连接到至少一个控制节点120a,b,它与外部IP网125(或其它数据通讯网)及存储区域网(SAN)130通讯。 Switch fabric 115a, b is also connected to the at least one control node 120a, b, it is an external IP network 125 (or other data communications network) and storage area network (SAN) 130 communications. 如远程执行的管理应用135能通过IP网125访问一个或多个控制节点,以帮助构筑平台100并部署虚拟化的PAN。 135 through the IP network management applications such as remote execution of 125 access one or more control nodes, to help build and deploy virtualization platform 100 of the PAN.

在某些实施例中,约24个处理节点105a-n,2个控制节点120,2个交换结构115a,b包含在单个机箱中并用固定的预先连线的点对点(PtP)链路网络互连。 In certain embodiments, 105a-n, 2 control node processing node 120, 2 to about 24 a switch fabric 115a, b and contained in a single chassis (PtP) link point fixed network interconnected with the prewired . 每个处理节点105是一块线路板,包括一个或多个(如4个)处理器106j-1,一个或多个网络接口卡(NIC)107,和本地存储器(如大于4G字节),它包括某些用于自举和初始化的BIOS固件及其他。 Each processing node 105 is a circuit board, comprising one or more (e.g., four) processors 106j-1, one or more network interface cards (NIC) 107, and local memory (e.g., greater than 4G bytes), it including certain BIOS firmware for bootstrapping and initialization and others. 对处理器106没有本地盘;相反,包括为分页所需存储器的所有存储器由SAN存储设备130处理。 No local disk processor 106; on the contrary, including the memory required for all memory pages processed by the SAN storage device 130.

每个控制节点120是一块线路板,包括一个和多个(如4个)处理器,本地存储器和本地磁盘存储,用于保持自举映象及初始文件系统的单独拷贝,后者用于自举未处理节点105和控制节点106所用的操作系统软件。 Each node 120 is a control board, and comprises a plurality (e.g., four) processor, a local memory and a local disk storage for holding a bootstrap image and a separate copy of the original file system, which is used from For untreated control node 105 and node 106 used by the operating system software. 每个控制节点经由连接到光纤通道链路122,124的100兆字节/秒光纤通道适配器卡128与SAN 130通讯,并经由具有连接到千兆位以太网链路121,123的一个或多个千兆位以太网NIC的外部网络接口129与因特网(或任何其它外部网)125通讯。 Each control node is connected via Fiber Channel links 122, 124 to 100 Mbytes / sec Fiber Channel adapter cards 128,130 communicate with the SAN, and having a via connected to a Gigabit Ethernet links 121, 123 or more Gigabit Ethernet NIC external network interface 129 to the Internet (or any other external network) 125 communications. (对SAN和外部网络连接可使用许多其它技术和硬件)。 (For SAN and external network connections and other techniques can be used a number of hardware). 每个控制节点包括低速以太网端口(未示出)作为专用管理端口,它不同于远程的基于Web的管理,可由管理应用程序135使用。 Each control node comprises a low speed Ethernet port (not shown) as a dedicated management port, which is different from the remote Web-based management, management application 135 can be used.

交换结构包括一个或多个如NIC-CLAN1000和Clan5300交换机的30个端口的千兆网交换机115,各种处理及控制节点使用相应的NIC与那样的结构模块通讯。 The switch fabric comprises one or more NIC-CLAN1000 and Clan5300 The switch 30 port gigabit switch 115, and controls various processing nodes using the corresponding communications module structure as the NIC. 千兆网交换结构具有非广播多重接入(NBMA-Non-BroadcastMultiple Access)网的语义。 Gigabit Ethernet switch fabric having a non-broadcast multiple access (NBMA-Non-BroadcastMultiple Access) Semantic Web. 所有节点间的通讯是经过交换结构。 All communication between nodes is through the switch fabric. 每个链路是作为NIC107和交换结构115中的端口之间的串行连接组成。 Each link is a serial connection between the switch fabric 115 and the connecting port NIC107 composition. 每个链路以112兆字节/秒的速率操作。 Each link operating at a rate of 112 Mbytes / sec.

在某些实施例中,多个机柜或机箱能互相连接以形成更大的平台。 In certain embodiments, a plurality of cabinet or chassis can be interconnected to form a larger platform. 在其它实施例中配置可以不同,例如能够消除冗余的连接,交换机和控制节点。 In other embodiments, the configuration may be different, for example, eliminate redundant connection, switch and control node.

在软件控制下,平台支持多重,同时和独立的处理区域网络(PAN)。 Under software control, the platform supports multiple, simultaneous and independent processing area network (PAN). 通过软件命令,每个PAN配置成具有对应的处理器106的子集,它能经过在PtP网络上模拟的虚拟局域网通讯。 By software command, each configured to have a PAN corresponding to the subset of the processor 106, virtual local area network can communicate through the PtP network simulation. 每个PAN也配置成具有对应的虚拟I/O的子系统。 Each PAN is also configured to have a corresponding virtual I / O subsystem. 不需要物理的部署或连线来建立PAN。 No physical connection to the deployment or build PAN. 在某些首选实施例中,在处理节点和/或控制节点上执行的软件逻辑模拟交换的以太网的语义;在处理节点和/或控制节点上执行的其他软件逻辑提供虚拟存储子系统功能,它遵循SCSI语义并对每PAN提供单独的I/O地址空间。 In certain preferred embodiments, software semantics Ethernet logic simulation executed on the processing node and / or nodes exchange control; other software logic executed on the processing node and / or control function node provides virtual storage subsystem, it follows a separate SCSI semantics and I / O address space of each PAN.

网络结构某些首选实施例允许管理员使用虚拟的组件,接口和连接来建立虚拟的,模拟的LAN。 The preferred embodiments of certain embodiments allows administrators to use a virtual network structure components, interfaces and to establish a virtual connection, analog LAN. 每个虚拟的LAN对平台100能是内部和专用的,或多个处理器能构成外部看作单个IP地址的处理器群集。 Each virtual LAN platform 100 and the internal energy is dedicated, or more processors can be configured as a single external IP address of the processor cluster.

在某些实施例中,那样建立的虚拟网络模拟交换以太网,尽管物理的底层网络是PtP网格。 In certain embodiments, virtual network set up as an analog switched Ethernet, although the underlying physical network is a PtP grid. 虚拟网络利用IEEE MAC地址,处理节点支持IEEE ARP处理以便用MAC地址识别和关联IP地址,因此,给定的处理器节点持续地回答ARP请求,无论该ARP请求是从平台内部或外部的节点来的。 The virtual network using the IEEE MAC address, the processing node supports the IEEE MAC address of the ARP processing for recognition and association with the IP address, therefore, the given processor node continuously answer ARP request, the ARP request is whether from inside or outside internet nodes of.

图2A示出示例性网络安排,它能被模型化或模拟。 FIG 2A shows an exemplary network arrangement, which can be modeled or simulated. 第一个子网202由处理节点PN1、PN2和PNk组成,它们能经过交换机206互相通讯。 First subnet 202 by the processing nodes PN1, PN2 and PNk composition, they can communicate with each other through the switch 206. 第二个子网204由处理节点PNk和PNm组成,它们通过交换机208互相通讯。 Second subnet 204 by a processing node and PNm PNk composition, which communicate with each other through the switch 208. 在交换以太网语义下,一个子网的一个节点能直接与该子网的其它节点通讯;例如,PN1能发送报文给PN2。 In switched Ethernet semantics of, a subnet node direct communication with other nodes of the subnet; e.g., PNl can send packets to PN2. 语义也允许一个节点与一组其它的节点通讯;例如PN1能发送广播报文到其它诸节点。 Semantic also allow a node to other nodes in the communication group; e.g. PN1 can send broadcast messages to all other nodes. 处理节点PN1和PN2不能直接与PNm通讯,因为PNm是在不同的子网。 PN1 and PN2 processing node can not communicate directly with PNm, because PNm are on different subnets. 为了PN1和PN2与PNm通讯,需要利用更高层的网络软件,那个软件对两个子网具有更充分的了解。 To PN1 and PN2 with PNm communications, need to use higher-level network software, the software has a fuller understanding of the two subnets. 虽然图中未示出,给定的交换机可以经过“上行链路”耦合到其它交换机等。 Although not shown, it may pass through a given switch "uplink" coupled to other switches. 下面将谈到,对那样上行链路的需要不同于当交换机是物理的时的需要。 As discussed below, the need for uplink is different from that when the switch is required when physical. 具体来说,因为交换机是虚拟的并在软件中模型化,它们能在水平方向按需要随意地扩展。 Specifically, because the switch is virtual and modeled in the software, they can expand freely as required in the horizontal direction. (相反,物理的交换机具有固定数目的物理端口,有时需要上行链路提供水平方向的伸缩性)。 (In contrast, the physical switch having a fixed number of physical ports, the uplink is sometimes necessary to provide flexibility in the horizontal direction).

图2B示出在某些实施例使用的示例性软件通讯路径和逻辑,以构建图2A的子网202和204。 2B illustrates an exemplary software and logical communication path used in certain embodiments of the embodiment, in order to construct the subnet 202 and 204 of FIG. 2A. 通讯路径212连接到处理节点PN1,PN2,PNk和PNm,具体说来,是它们对应的处理器端的网络通讯逻辑210,它们也将处理节点连接到控制节点。 Communication path 212 is connected to the processing node PN1, PN2, PNk PNm and, specifically, their corresponding end processor network communication logic 210, which will be connected to the control node processing node. (虽然为明确起见画成逻辑的单个实例,PNk可具有对应处理器逻辑的多个实例,例如每个子网一个)。 (For clarity, although a single instance logic painted, PNK may have a plurality of instances corresponding to the logical processors, such as a per subnet). 在首选实施例中,管理逻辑和控制节点逻辑负责建立,管理和断开通讯路径。 Embodiment, the management logic, and logic control node responsible for the establishment, management, and disconnect the communication path in the preferred embodiment. 不允许单个处理节点建立那样的路径。 It does not allow a single processing node to establish such routes.

下面将细述,处理器逻辑和控制节点逻辑一起在那样通讯路径上模拟交换以太网的语义。 Detailing below, processor logic, and analog control of logical nodes in a semantic switched Ethernet as the communication path. 例如,控制节点具有模拟某些(不必是全部)以太网交换的语义的控制节点端的虚拟交换逻辑214,而处理器逻辑包括模拟某些(不必是全部)以太网驱动器的语义的逻辑。 For example, some with analog control node (not necessarily all) of the Ethernet switch the control node side semantic virtual switching logic 214, and the processor includes an analog logic some (not necessarily all) of the logical semantics of the Ethernet driver.

在一个子网中,一个处理器节点经过相应的虚拟接口212直接与其它节点通讯。 In one subnet, a processor node 212 to communicate directly with other nodes via the corresponding virtual interface. 类似地,处理器节点能经过单独的虚拟接口与控制节点逻辑通讯。 Similarly, the node processor can communicate with the virtual interface via a separate control node logic. 在某些实施例中,底层交换结构和相关逻辑(如交换结构管理器逻辑,未示出)提供在点对点的网格上建立和管理那样的虚拟接口(VI)的能力。 In certain embodiments, the switch fabric and the underlying associated logic (e.g., switch fabric management logic, not shown) and the ability to create a virtual as in the management of the grid point to point interfaces (VI),. 此外,这些虚拟接口能以可靠冗余的方式建立,这里称为RVI。 In addition, these virtual interfaces to establish a reliable redundant way, here referred to as RVI. 在此叙述中,术语虚拟接口(VI)和可靠的虚拟接口(RVI)可互换使用,同为在VI和RVI之间的选择主要取决于以系统资源为代价系统的所希望的可靠性的程度。 In this description, the term virtual interface (VI) and reliable virtual interface (RVI) used interchangeably, with the selection between VI and RVI depends mainly on the cost of the system resources to system reliability desired degree.

联合地参考图2A-B,若节点PN1与节点PN2通讯,正常地地通过虚拟接口2121-2。 Referring jointly to FIGS. 2A-B, if the communication node PN1 and PN2 node 2121-2 via the normal virtual interface. 但是若如VI2121-2并非运行良好,首选实施例允许在PN1和PN2之间的通讯经过交换模拟逻辑进行。 However, if such VI2121-2 not run well, the preferred embodiment allows communication between PN1 and PN2 is exchanged via analog logic. 在此情况,报文可经VI2121-switch206和VI212switch206-2发送。 In this case, the message may be sent via the VI2121-switch206 and VI212switch206-2. 若PN1广播或多播报文到子网202中其它诸节点,它是经过虚拟接口2121-switch206发送报文到控制节点端的逻辑214。 If PN1 broadcast or multicast packets to all other nodes in the subnetwork 202, which is the result of the virtual interface 2121-switch206 send packets to the end node 214 of the control logic. 然后控制节点端逻辑214通过使用对应的VI克隆和发送报文到其它对应的节点来模拟广播或多播功能。 Node logic 214 then controls the terminal by using the corresponding VI Cloning and send packets to other nodes corresponding to the analog broadcast or multicast function. 可使用同样或类似的VI传递其它需要控制节点端逻辑的报文。 Using the same or similar transfer VI other desired nodes of the logic control packets. 例如如下所述,控制节点端逻辑包括支持地址解析协议(ARP)的逻辑,而VI被用于传送ARP回答和请求到控制节点。 The following example, the control logic includes a logical end node supporting address resolution protocol (ARP), and VI is used to transmit an ARP reply and the request to the control node. 虽然上面描述在处理器逻辑和控制逻辑之间只提出一个VI,许多实施例使用若干那样的连接。 While the above description sets forth only a VI between the processor logic and control logic, such as a number of embodiments using several connections. 此外,虽然图中在软件通讯路径中建议对称性,结构实际上允许非对称的通讯,例如如下讨论,对群集通讯服务,数据包经控制节点路由。 In addition, although FIG suggested Symmetry software communication path, an asymmetrical structure actually allow communications, for example as discussed below, a cluster communication service, via the control node routing packets. 但返回通讯能在节点之间直接进行。 But the return can be direct communication between the nodes.

注意如图2A的网络,没有在节点PN2和PNm之间通讯的机制。 Note that the network in FIG. 2A, there is no mechanism for communication between nodes and PN2 PNm. 而且,通过集中管理和建立(不是通过处理节点)通讯路径,那样的路径不能由处理节点建立,且已定义的子网连接不能被处理器破坏。 Also,, as the path can not be established by the processing node through centralized management and building (not by processing node) communication path, and the defined subnetwork connection can not be destroyed processor.

图2C示出为实现图2A和2B的子网的某些实施例的示例性物理连接。 FIG 2A and FIG. 2C illustrates an exemplary physical embodiment of certain embodiments of the subnet connected 2B is achieved. 具体说来,处理连网逻辑210的每个实例通过互连110的PtP链路216与交换结构115通讯。 Specifically, each instance of networked processing logic 210 via link 216 PtP communication switching structure 115 and the interconnect 110. 类似地,控制节点具有交换逻辑214的多个实例,每个经PtP连接216连接到交换结构。 Similarly, the control node having a plurality of instances of switching logic 214, each warp PtP connections 216 connected to the switching structure. 图2B的虚拟接口包括在这下物理链路上传递信息的逻辑,下面将进一步描述。 FIG. 2B virtual interface logic comprises transmitting the information over this physical link, as described further below.

为建立和配置那样的网络,管理员定义PAN的网络拓扑结构并规定(如通过在管理软件135中的实用程序)各种节点的MAC地址分配。 For establishing and configuring such networks, the network administrator defines the topology and PAN predetermined MAC address (e.g., utility by the management software 135) of the dispensing various nodes. MAC地址是虚拟的,标识虚拟接口,且与任何指定物理节点无关。 MAC address is virtual, the virtual interface identifier, and independent of any specific physical node. 在某些实施例中,MAC地址遵循IEEE的48位地址格式,但其中内容包括“本地管理”(设置成1),最初在其上定义的虚拟接口的控制界定120的序号(较下面),和从在控制节点上的永久性顺序计数器来的计数值,它被保持在控制节点的NVRAM中。 In certain embodiments, MAC address, an IEEE 48-bit address format, but includes "local administration" (set to 1), define the initial number of 120 (more below) to control the virtual interface defined thereon, and permanently to the order on the control node count value, which is held in the NVRAM control node. 这下MAC将用于在第二层上识别节点(如传统那样)。 This is the MAC will be used to identify the node on the second layer (as is conventional). 例如,在回答ARP请求(无论是从内部节点到PN,或在外部网络上)时,这些MAC将包括在ARP回答中。 For example, in answer ARP requests (either from the PN to the internal node, or on the external network), the MAC will be included in the ARP reply.

控制节点端的连网逻辑维护数据结构,它包含反映LAN的连接的信息(例如,哪个节点可以与哪个节点通讯)。 Logic control terminal networked node maintains a data structure that contains information reflecting the LAN connection (e.g., which may be associated with which node communication node). 控制节点逻辑还分配和指定对已定义的MAC地址的VI(或RVI)映射,并分配和指定在控制节点之间以及在控制节点和处理节点之间的VI(或RVI)。 Logic and control node also allocates the MAC address designated VI defined (or RVI) mapping, and partitioned between control and specify the nodes and VI (or RVI) between the control node and the processing node. 在图2A的例子中,逻辑将分配和指定图2B的VI212。 In the example of FIG. 2A, the logic and the assignment of Figure 2B designated VI212. (在某些实施例中的VI和RVI的命令是使用的交换结构及交换结构的管理器逻辑的结果)。 (VI RVI and commands in some embodiments is the result manager logic switching fabric switching structure used).

当每个处理器自举时,基于BIOS的自举逻辑初始化节点105的每个处理器106,除其他工作外还建立(或发现)到控制节点逻辑的VI212。 When each processor bootstrapping, based BIOS bootstrap node initialization logic 105 of each processor 106, among other work has established (or discover) to the control node logic VI212. 然后处理器节点从控制节点获得相应的数据链路信息,如处理器节点的MAC地址,和在同一数据链路结构中其它设备的MAC标识。 The processor node obtains the corresponding data link information from the control node, such as the MAC address of the processor node, and identifying other MAC devices on the same data link structure. 然后每个处理器用控制节点注册其IP地址,然后将IP地址与节点以及RVI(如注册到达的RVI)联系起来。 Each processor then registers its IP address with the control node, and the IP address of the node and the RVI (registered as arriving RVI) link. 以此方式,控制节点能够将在一个子网中对每个节点的每个虚拟MAC的各IP地址联系起来。 In this way, the control node can be linked to each IP address for each virtual MAC each node in a subnet. 除此之外,处理器节点还获得对它到其它节点或到控制节点连网逻辑的连接的RVI,或VI有关的信息。 In addition, it also received a processor node to other nodes or to a control node connected RVI networking logic, or information concerning VI.

这样,在自举及初始化之后,各种处理器节点将掌握它们的第二层的数据链路连接。 Thus, after initialization and bootstrapping, a variety of processor nodes to their master data link layer of the second connector. 下面将解释,在处理器正常处理过程中,作为地址解析协议的结果确定了第三层(IP)的连接,尤其是第三层到第二层的相关性。 It will be explained below, the processor in the normal processing, as a result of ARP to determine the connection of the third layer (IP), in particular relevance to the second layer, the third layer.

图3A细化了某些实施例处理器端连网逻辑210,图3B细化了某些实施例的控制节点端连网逻辑310。 FIG 3A refine certain embodiments networking logic processor end 210 embodiment, FIG. 3B refined some control logic networked node side 310 of the embodiment. 处理器端逻辑210包括IP堆栈305,虚拟网络驱动器310,ARP逻辑350,RCLAN层315,和冗余千兆网驱动器320a,b,控制节点端逻辑310包括冗余千兆网驱动器325a,b,RCLAN层330,虚拟群集代理逻辑360,虚拟LAN服务器335,ARP服务器逻辑335,虚拟LAN代理340,和物理的LAN驱动器345。 Processor logic 210 includes a terminal IP stack 305, virtual network driver 310, the ARP logic 350, RCLAN layer 315, and the redundant gigabit Ethernet drivers 320a, b, the control node of the logic 310 includes a redundant gigabit Ethernet drivers 325a, b, RCLAN layer 330 in virtual cluster proxy logic 360, the virtual LAN server 335, the ARP server logic 335, the virtual LAN agent 340, and the physical LAN driver 345.

IP堆栈IP堆栈305是用由处理节点106使用的操作系统(如Linux)提供的通讯协议堆栈。 IP stack IP stack 305 is the operating system (such as Linux) for use by a processing node 106 provides a communication protocol stack. IP堆栈为在处理器106上执行的应用程序及操作系统提供第三层的接口,以便与模拟的以太网通讯。 IP stack for the application executing on the processor 106 and operating system interface layer so as to simulate the Ethernet communications. IP堆栈结合将第三层IP地址作为信息包的目的地,向虚拟以太网层次310提供该信息包。 Binding third layer IP stack as the destination IP address of the packet, the virtual Ethernet level 310 provides the packet. IP堆栈逻辑是传统的,不同处在于某些实施例免去校验和计算及逻辑。 IP stack logic is conventional except that the embodiment eliminates the checksum calculation logic and certain embodiments.

虚拟以太网驱动器虚拟以太网驱动器310对IP堆栈305看来象“真实的”以太网驱动器。 Virtual Ethernet driver virtual Ethernet driver 310,305 seems like a "real" Ethernet driver for the IP stack. 在这方面,虚拟以太网驱动器310从IP堆栈接收IP包或数据包,用于随后在网上发送,且它从网络接收要作为IP包提交给堆栈的包信息。 In this regard, virtual Ethernet driver 310 receives the IP packet from the IP stack or data packets for subsequent transmission on the Internet, and it receives a packet of information to be submitted to the stack as IP packets from the network.

堆栈建立MAC标题。 Stack establish MAC title. 可使用在堆栈中的“正常”的以太网码。 The stack can be used in "normal" Ethernet code. 虚拟的以太网驱动器接收带有已建立的MAC标题的包和已经在标题中的正确的MAC地址。 Virtual Ethernet driver receives the packet and has the correct MAC address with the MAC title established in the title.

参考图4A-C,在实质性部分中虚拟以太网驱动器310在405将进行的IP数据包出队,使得包能在网络上发送。 With reference to FIGS. 4A-C, in a substantial portion of the virtual Ethernet driver 310 in the IP packet 405 will be dequeued so that packets can be sent over the network. 使用标准的IP堆栈ARP逻辑。 ARP IP stack using standard logic. 下面将解释,驱动器拦截所有进入和离开系统的ARP包,修改它们,使得合适的信息出现在每个节点的ARP表中。 Will be explained below, the drive system intercepts all entering and exiting the ARP packet, modify them, so that the appropriate message in the ARP table of each node. 在数据包排队等待以太网驱动器之前,正常的ARP逻辑将正确的MAC地址放在输出的包的链路层标题中。 Link layer header packet before the packet queued for Ethernet driver, the normal logic ARP MAC address in the correct output. 然后驱动器检查链路层标题及目标MAC以确定如何发送该包。 The drive then checks the destination MAC and Link layer header to determine how to send the packet. 驱动器不直接处理ARP表(除非有ARP条目的偶然的不合法性)。 It does not deal directly drive the ARP (ARP entry unless there is no chance legitimacy).

驱动器310在415确定,在出队的包中ARP逻辑350是否具有与IP地址相关的MAC地址信息(下面详述)。 415 in the driver 310 is determined, ARP logic 350 whether the MAC address information associated with the IP address (described below) dequeuing package. 若ARP逻辑350具有该信息,于是在420就使用该信息发送包。 If the ARP logic 350 has this information, it then uses this information in a transmission packet 420. 若ARP逻辑350不具有该信息,驱动器需要确定那样的信息,在某些首选实施例中,如结合图4B-C讨论的那样,该信息作为实现ARP协议的结果而获得。 If the ARP logic 350 does not have this information, the driver needs to determine that information, in certain preferred embodiments, as described in FIG. 4B-C, as discussed above, the information achieved as a result of the ARP obtained.

若ARP逻辑350具有MAC地址信息,驱动器分析从ARP逻辑350返回的信息以确定何处及如何发送该包。 If the ARP logic 350 having the MAC address information, the drive logic 350 analyzes information from the returned ARP to determine where and how to send the packet. 具体说来,驱动器查看地址以判断MAC地址是否为合法格式或特定非法格式。 Specifically, the drive to view the address to determine whether the MAC address format is legal or illegal particular format. 例如,在一个实施例中,通过在MAC地址的第一字节中设置本地管理位,多播位,和另外预定位模式的组合发信号给内部节点(即平台的内部PAN节点)。 For example, in one embodiment, by providing the local management bit in the first byte of the MAC address, multicast bit, and combinations additional signaling bit pattern to a predetermined internal node (i.e., node platform internal PAN). 包罗万象的模式是极不可能为合法的模式。 All-inclusive model is highly unlikely that a legitimate mode.

若从ARP逻辑返回的ARP地址为合法的格式,与MAC地址相关的IP地址是对至少相关子网为外部的节点,且在首选实施例中对平台是外部的。 If the address from the ARP ARP logic returns to valid format, IP address, MAC address is associated with the external nodes of at least the subnet, and, in the preferred embodiment it is external to the platform. 为提交那样的包,驱动器将TLV(类型-长度-值)的标题预先挂在包上。 Such packages for the submission, the drive TLV (Type - Length -) of the title previously hung on the package. 然后逻辑在预先建立的VI上发送包到控制节点。 And VI on the pre-established logic transmits the packet to the control node. 然后控制节点恰当地处理余下的发送工作。 Then the control node transmitting appropriately deal with the remaining work.

若从ARP逻辑350返回的MAC地址信息为特定的非法格式的,该非法格式表明该IP地址的节点是对内部节点的,在MAC地址信息中的信息被用于帮助识别直接连接两个处理节点的VI(或RVI)。 If the returned ARP logic 350 MAC address information, the illegal format indicates a specific format of illegal IP address of the node it is an internal node, the information in the MAC address information is used to help identify the direct connection of two processing nodes the VI (or RVI). 例如,ARP表的条目能保持识别RVI212的信息,用于将包(如2121-2)发送到另外的处理节点。 For example, entries in the ARP table to keep information identifying RVI212 for the packet (e.g. 2121-2) sent to another processing node. 驱动器将TLV标题预先挂在包上。 The pre-driver hung on TLV packet header. 然后将地址信息以及识别以太网协议类型的信息放在标题中。 Address information, and then identify the Ethernet protocol type of information in the header. 然后逻辑选择发送封装的包的合适的VI(或RVI)。 Select logic then sends the encapsulated packet to the appropriate VI (or RVI). 若该VI(或RVI)运行良好,它被用于运载该包;若它并不运行良好,该包被发送到控制节点交换逻辑(下面详述),使得交换逻辑将其发送到适当的节点。 If the VI (or RVI) works well, which is used to deliver the packet; if it does not work well, the packet is sent to the control node switching logic (described below), so that the switching logic will send it to the appropriate node . 虽然ARP表可包含实际指定要用的RVI的信息,可以利用许多其它技术。 Although the ARP table may contain information specified RVI actual use, the number of other techniques may be utilized. 例如,表中的信息可以间接地提供那样的信息,如通过指向关心的信息或通过识别关心的信息,即使表不包含该信息。 For example, the information table may provide that information indirectly, such as by pointing information of interest or by the identification information of interest, even if the table does not contain this information.

对任何多播或广播类型的报文,驱动器发送报文到已定义的VI上的控制节点。 Any type of broadcast or multicast packets, packets sent to the drive control node defined on the VI. 然后控制节点相应地克隆该包,并将其发送到所有节点(除发送的节点以外),因而到上行链路。 The packet control node is then cloned accordingly, and sends it to all nodes in the uplink (other than the transmission node), thereby to.

若不存在ARP映射,则更高的层次永不会将包发送到该驱动器。 If the presence of the ARP, the higher level packets will never be sent to the driver. 若不存在可用的数据链路层映射,在ARP转换完成前该包被置于一旁。 If there is a data link layer mapping is available, the conversion is completed before the ARP packet is set aside. 一旦ARP层次结束ARP工作,挂有ARP的包使建立它们的数据链路标题,且该包随后被送到驱动器。 Once the level of ARP ARP working end, hanging ARP packets so that they establish data link header, and the packet is then sent to the driver.

若对于IP包的IP地址,ARP逻辑没有从IP堆栈的映射,结果驱动器310不能确定有关的地址信息(即MAC地址或RVI相关信息),驱动器通过遵循ARP协议获得那样的信息。 If the IP address for the IP packet, there is no ARP logical mapping from IP stack driver 310 can not determine the result of the address information (i.e., MAC address, or RVI-related information), the driver by following the ARP protocol to obtain that information. 参考图4B-C,驱动器在425建立包含有关IP地址的ARP请求包,对此IP地址在本地ARP表中没有MAC映射。 Referring to FIG 4B-C, the driver contains information about the establishment of the IP address in the ARP request packet 425, the IP address for this MAC is not in the local ARP table mapping. 然后节点在430预先将TLV-型标题挂到ARP包上。 Then the node 430 is previously linked to the headline TLV- ARP packet. 然后ARP请求经过专用的RVI被送到控制节点端的连网逻辑-具体说来是虚拟LAN服务器335。 Then RVI ARP request is sent through a dedicated control logical node side networking - specifically the server 335 is a virtual LAN.

下面将详述,ARP请求包在435由控制节点处理,并在440广播到有关节点。 Detailed below, the ARP request packet from the control node 435 in the process 440 and broadcast to the node concerned. 例如,控制节点将标记出,请求节点是否为IP服务群集的部分。 For example, the control node will be marked, whether the node is part of an IP service request cluster.

在有关节点的以太网驱动器逻辑310在445接收ARP回答,并在450借助对节点的IP堆栈作出调用,通过将目标IP地址与本地配置的IP地址的列表作比较确定,它是否为ARP请求的目标。 Ethernet driver logic about nodes 310 445 receiving ARP reply, and means to make 450 node's IP stack invoked by listing the destination IP address of the local configured IP address for comparison to determine whether it is the ARP request aims. 若它不是目标,它拒绝该包而不作修改。 If it is not a goal, it rejects the packet without modification. 若它是目标,驱动器在460从TLV标题建立本地MAC标题,并在465更新ARP表而且建立ARP回答。 If it is the goal of establishing a local drive in 460 TLV MAC title from the title, and update the ARP table 465 and establish ARP reply. 驱动器修改在ARP请求中的信息(主要是源MAC),并随后通常将ARP请求上传,用于更高层次的处理。 Drive modify information in the ARP request (mainly the MAC source), and then upload the ARP request is typically, for higher level processing. 正是该更高的层次在必要时形成ARP回答。 It is this higher level ARP reply form if necessary. 回答包含回答节点的MAC地址及其它,并具有在TLV标题中设置的位,指出回答是从本地节点来。 Answer answers contains the MAC address and other nodes, and having a bit set in TLV title, the answer is to be noted from the local node. 在这方面,节点按照IETF-类型的ARP语义响应(与ATM ARP协议不同,在后者中ARP回答被集中处理)。 In this regard, the nodes in the semantic response IETF- ARP type (with different ATM ARP protocol, in which the focus is an ARP reply). 然后在470回答被发送。 Answer at 470 is then transmitted.

下面将更详细描述,控制节点逻辑335在473接收回答并修改它。 Described in more detail below, the control node logic 335 receives the answer 473 and modify it. 例如,控制节点可以用识别源机柜的信息,处理节点号,RVI连接号,通道,虚拟接口号,和虚拟LAN的名称替代回答的内部节点的MAC地址。 For example, the control node may use the information identifying the source of the cabinet, the processing node number, the RVI connection number, the channel, the virtual interface number, and the name of the virtual LAN MAC address of the internal node of the alternative answers. 一旦ARP回答被修改,控制节点逻辑在475随后发送ARP回答给适当的节点,即发送该ARP请求的节点,或在特定实例中送到IP服务群集的负载均衡器,这将在下面讨论。 Once the ARP reply is modified, the control logic node then sends an ARP reply 475 to the appropriate node, i.e. the node transmits the ARP request, to the load balancer or IP services in a particular instance of the cluster, which will be discussed below.

封装的ARP回答最终在480被接收。 ARP reply package 480 is finally received. 若回答节点是外部节点,ARP回到包含回答节点的MAC地址。 If the answer node is an external node, ARP back to the MAC address of the node that contains the answer. 若回答节点是内部节点,替代地,ARP回答包含识别有关RVI的信息以便与节点通讯。 If the answer node is an internal node, alternatively, to answer the ARP contains identification information about the communication node so as RVI. 在两种情况在485更新本地表。 Update local table 485 in both cases.

挂上的数据包在487出队,且在493选择合适的RVI。 Put in dequeued packets 487, 493 and suitable selection RVI. 如上讨论,合适的RVI根据目标节点是内部或外部来选择。 As discussed above, suitable RVI is selected according to the target node is an internal or external. TLV标题被预先挂到数据包并在495被发送。 TLV titles are linked to the previously transmitted data packet and is 495.

为了在虚拟LAN中通讯,最大发送单位(MTU)被配置成16896字节。 To the virtual LAN communication, the maximum transmission unit (MTU) is configured to 16896 bytes. 即使配置的MTU是16896字节,在包被送到外部网络时以太网驱动器310予以辨别。 Even MTU configuration is 16896 bytes, to identify when the package 310 to be sent to the external network is an Ethernet driver. 通过使用路径MTU发现,ICMP和IP堆栈改变,路径MTU在源节点105处被改变。 By using the Path MTU Discovery, ICMP, and IP stack change, the path MTU is changed at the source node 105. 此机制也被用于触发包的校验和。 This mechanism is also used to trigger packet checksum.

本发明的某些实施例通过在虚拟LAN服务器335处和在虚拟LAN驱动器310中的逻辑的组合支持混合模式。 Certain embodiments of the present invention, by a virtual LAN server 335 and logic in the virtual LAN driver 310 supports a combination of mixed mode. 当虚拟LAN驱动器310从虚拟LAN服务器335接收混合模式的报文时,报文包含有关识别希望进入混合模式的接收器的信息。 When the virtual LAN driver 335 receives packets 310 from the mixed-mode virtual LAN server, a message including information about the identification of the receiver into the desired hybrid mode. 此信息包括接收器的位置(机柜,节点等),在接收器上混合虚拟接口310的接口号(对信号分离包所需要的),和接收器所属的虚拟LAN的名称。 This information includes the position of the receiver (cabinets, nodes, etc.), the name of the receiver in the hybrid interface number 310 of the virtual interface (packet signal separation required), and a receiver belongs to the virtual LAN. 此信息随后被驱动器310用于确定,如何发送混合的包到接收器(哪个RVI,或用于发送包的其它机制)。 This information is then used to determine the drive 310, how to send the mixed packet to the receiver (which the RVI, or other mechanism for transmitting packets). 虚拟接口310在同一虚拟LAN上维护混合的监听者的表。 Virtual table interface 310 to maintain listener mixed on the same virtual LAN. 当发送节点接收混合模式的报文时,它将随之更新它的混合表。 When the transmitting node receives a mixed mode packet, mixing it updated its table.

当包在虚拟以太网驱动器上发送时,此表被审查。 When the packet is sent on a virtual Ethernet driver, the table is reviewed. 若表非空,则虚拟以太网接口310将做下列工作:·若发出的包是广播或多播的,将不发送混合拷贝,正常的广播操作将发送包到混合的监听者·若包是其目标不是混合的监听者的单播包,包将被克隆并送到混合的监听者。 If the table is not empty, the virtual Ethernet interface 310 will do the following: • If the package is sent by broadcast or multicast will not send mixed copies, normal operation of broadcast packets will be sent to the mixing listeners • If package is listener's goal is not mixed unicast packet, the packet will be cloned to the mixing and listener.

标题TLV包括目标用来信号分离并校验输入的包的额外信息。 TLV header includes a target for signal separation and additional parity packets input. 此信息的部分是目标的虚拟以太网接口号(在接收节点的目标设备号)。 This information is part of the target virtual Ethernet interface number (device number in the target receiving node). 因为在实际的包目标及混合的目标之间这些可能是不同的,此标题不能简单地被克隆。 Because the packet between the actual target and mixing these goals may be different, the title can not be cloned easily. 因此,为了每个包克隆到每个混合监听者,必须对每个标题分配存储器。 Therefore, to each packet to each mixing clone listener, the memory must be allocated to each title. 在对混合包建立包标题时,必须设置包类型,指出该包是混合发送而不是单播发送。 In establishing the packet header mixed bag, you must set the packet type, indicating that the packet is sent instead of mixed unicast transmission.

虚拟以太网驱动器310也负责处理冗余的控制节点连接。 Virtual Ethernet driver 310 is also responsible for handling redundant control nodes. 例如,虚拟以太网驱动器通过发送脉动信息包TLV到每个连接的RVI,周期地测试端到端的连接。 For example, virtual Ethernet driver TLV RVI connected to each packet by sending a pulsating periodic testing end to end connection. 这使虚拟以太网驱动器能确定节点是否停止响应,或停止的节点是否开始再次响应。 This allows the virtual Ethernet driver can determine if the node stops responding, or whether to stop the node start responding again. 当一个RVI或控制节点120被判断停止了,以太网驱动器将通过活动的控制节点发送通信量。 When a control node 120 or RVI is judged stopped, the Ethernet driver by the control node transmits the traffic event. 若两个控制节点在工作,驱动器310试图在两个节点之间负载均衡通信量。 If both the control node on the drive 310 between the two nodes attempt to load balance traffic.

本发明的某些实施例提供性能改善。 Certain embodiments of the present invention provide improved performance. 例如,带着对IP堆栈305的修改,只在平台100内发送的包不作校验和,因为平台100的所有单元提供错误检测并保证数据的提交。 For example, with modifications to the IP stack 305, only the packet transmitted within the platform 100 and the checksum is not, because all platforms unit 100 is provided to ensure that the error detection and data submitted.

此外,为了在PAN中(或甚至在平台100中)通讯,RVI能配置成使得包大于以太网允许的最大尺寸。 Further, in order in the PAN (100 or even the internet) communication, can be configured such that the RVI Ethernet packets larger than the maximum size allowed. 因此,虽然在某些实施例中模型以太网行为,最大包尺寸可以被破坏以改善性能。 Thus, although in this embodiment the model behavior of Ethernet in certain embodiments, the maximum packet size to improve performance may be damaged. 作为数据链路层的部分,实际的包尺寸可以协商确定。 As part of the data link layer, the actual packet size can be negotiated.

控制节点的故障或者由来自RCLAN层的通知或者由脉动信息包TLV的故障检测。 The control node failure or a failure detection notification from RCLAN layer or pulsations of TLV packet. 若一个控制节点失败,以太网驱动器310将通信量只送到余下的控制节点。 If a control node fails, the Ethernet driver 310 to the rest of the traffic control node only. 以太网驱动器310将通过来自RCLAN层的通知或脉动信息量TLV的恢复辨别控制节点的恢复。 Ethernet driver 310 will be restored by restoring the control node notice or pulse discriminating information from TLV RCLAN layer. 一旦控制节点恢复,以太网驱动器310将恢复负载均衡。 Once the control node recovery, Ethernet driver 310 will resume load balancing.

若一节点检测到它不能通过直接RVI(如上描述的)与另外节点通讯,该节点试图经过起交换器作用的控制节点通讯。 If a node detects that it can not be directly the RVI (as described above) with a further communication node, the node attempts from the control node through the communications switch acting. 那样的故障能由较低的RCLAN层根据如不能收到虚拟接口的确认,或通过脉动信息机制检测的故障发信号通知。 As faults are from a lower layer according RCLAN not receive acknowledgment as virtual interfaces, or detected fault signaled by pulsation mechanisms. 在此情况,驱动器随即标注在TLV标题中的位,指出该报文是单播,并发送包到控制节点,使得它能发送包到希望的节点(例如若需要时根据IP地址)。 In this case, the driver then labeled TLV header bits indicating the packet is unicast, and transmits the packet to the control node, so that it can send the packet to a desired node (e.g., by IP address if needed).

RCLAN层RCLAN层315负责处理冗余互连NIC107的冗余性,故障在线恢复和负载均衡逻辑。 RCLAN layer RCLAN redundant interconnection layer 315 is responsible for processing redundancy, load balancing and fault recovery line of logic NIC107. 这包括检测故障,在故障时在冗余的连接上再路由通信量,负载均衡,和不能提交通信量的报告回到虚拟网驱动器310。 This includes detecting a fault, when the fault re-route traffic over redundant connections, load balancing, and reports back to the traffic can not submit a virtual network driver 310. 当在任何RVI上存在使RVI不能使用的致命性错误或若任何RVI由于任何理由停止时,虚拟以太网驱动器310期望能够异步地得到通知。 When there is any RVI RVI make fatal errors can not be used or if any RVI due to any reason to stop, virtual Ethernet driver 310 can expect to be notified asynchronously.

在正常情况,在每个处理器上的虚拟网络驱动器310试图在可用的控制节点之间负载均衡送出的包。 In normal circumstances, the virtual network driver in each processor 310 attempts to load balancing packets sent between the available control node. 这能通过在可用的控制节点之间的简单循环更迭做到,或通过持续跟踪在每个控制节点有多少字节已被发送,并总是通过发送最少字节的控制节点发送来做到。 This can be a simple loop between the control node is available to do the change, or By tracking how many bytes have been sent, and to do always sent by sending at least the control node of each byte of the control node.

RCLAN提供在核心之间高的带宽(每条路径224MB/秒)、低延时、可靠的异步点对点通讯。 RCLAN provided (each path 224MB / sec), low latency and reliable high-end communication between the asynchronous core bandwidth. 若作出了最大努力提交数据,数据未能被提交,通知数据的发送者。 If the best efforts to submit data, data is not submitted, notify the sender of the data. RCLAN使用二个Giganett Clan1000卡提供在核心之间的冗余通讯路径。 RCLAN use two Giganett Clan1000 cards provide redundant communication paths between the cores. 它无缝地恢复在clan1000卡或千兆网交换机中的单个故障。 It seamlessly recover a single fault in clan1000 Gigabit Ethernet card or switch. 它检测丢失的数据和数据错误,并在需要时重发送数据。 It detects data errors and missing data, and retransmission data when needed. 只要一个连接在部分地工作,例如错误率不超过5%,通信将不会中断。 As long as a connection part to work, for example, the error rate does not exceed 5%, the communication will not be interrupted. RCLAN的客户包括RPC机制,远程SCSI机制,和远程以太网。 RCLAN clients include RPC mechanism, remote SCSI mechanism, and remote Ethernet. RCLAN还提供流控制的简单形式。 RCLAN also provides a simple form of flow control. 低的延时和高的并发性通过允许对每个设备的多个同时请求由处理器节点发送到控制节点来实现,使得这些请求能尽可能快地传输到设备,或另选地使得与在处理器节点上排队所有请求相反,它们排队在尽可能地靠近设备处等待完成。 Low latency and high concurrency by allowing each of the plurality of devices simultaneously request to the control node by the node processor to achieve, so that they can transmit the request to the device as quickly as possible, or alternatively be such that the Instead all queued requests on a processor nodes, which are as close to the line at the apparatus waits for completion.

在控制节点端的RCLAN层330类似于上述地操作。 In the control node side RCLAN layer 330 similar to the operation of the above-described manner.

千兆网驱动器千兆网驱动器逻辑320是负责为千兆网NIC107提供接口的逻辑。 Gigabit Ethernet Gigabit Ethernet driver drive logic 320 is responsible for providing interfaces for Gigabit Ethernet NIC107 logic. 无论它在处理器106上或控制节点120上。 120 on whether it on the processor 106 or control nodes. 简而言之,千兆网驱动器逻辑建立与VIid相关的VI连接,使得如RCLAN315和以太网驱动器310那样的较高层只需要理解VI的语义。 Briefly, Gigabit Ethernet driver logic associated with establishing the connection VIid of VI, such as RCLAN315 Ethernet driver 310 and higher layers need only understand the semantics as the VI.

千兆网驱动器逻辑负责在每个节点为缓冲器及VI的队列分配存储器,用于检查NIC107以便知道其连接和存储器分配。 Gigabit Ethernet driver logic at each node is responsible for allocating memory buffer queue and VI for checking NIC107 in order to know their connections and memory allocation. 某些实施例使用由千兆网驱动器提供的VI连接。 Certain embodiments provided by the VI Gigabit Ethernet drive. 千兆网NIC驱动器码建立虚拟接口对(即VI),并将其赋予对应的虚拟接口id。 Gigabit Ethernet NIC driver code to establish a virtual interface (i.e. VI), and gives the corresponding virtual interface id.

每个VI是在一个千兆网端口和另一个之间建立的双向连接,或更确切地说是在一个节点上的存储器缓冲器及存储器队列和另一个上的缓冲器及队列之间的连接。 Each VI is two-way connection between a Gigabit Ethernet port and another established, or rather in the buffer memory connected between a node and a memory buffer and a queue on a queue and the other . 如上所述,端口和存储器的分配由NIC驱动器处理。 As described above distribution, and a memory port handled by the NIC driver. 通过将数据放入NIC知道的缓冲器并借助将其写到特定的存储器映射寄存器而触发动作,来将数据发送出去。 Triggered by the operation data into a buffer and with the NIC know to write a particular memory-mapped registers, to send data out. 在接收端,数据出现在缓冲器而完成状态出现在队列中。 At the receiving end, the data appears in the buffer in the queue is completed state occurs. 若发送和接收程序能够在连接的缓冲区中产生和使用报文,数据决不需要拷贝。 If the program can send and receive messages generated and used in the connection of the buffer, data will not need to copy. 若操作系统能将连接的缓冲器及控制寄存器“存储器-映射“到应用程序的地址空间,发送甚至能直接从应用程序到应用程序。 If the operating system can control the buffer register and "Memory - Mapping" connected to the address space of the application, or even can be directly transmitted from the application to the application. 每个千兆网端口能在其上支持1024个同时的VI连接,并用硬件保护互相分开地保存它们,所以操作系统和完全不同的应用程序能完全地共享单个端口。 Each port can support a Gigabit Ethernet connection 1024 VI simultaneously, and save them separately from each other with hardware protection, the operating system and applications completely completely different share on its single port. 在本发明的一个实施例中可以同时建立14个从每个端口到每个另外的端口的VI连接。 In one embodiment of the present invention may be established from each port 14 is connected to each of the other port simultaneously VI.

在首选实施例中,NIC驱动器以冗余对方式建立VI连接,该对的一个连接通过两个交换结构115a,b的一个而另一个通过另外的交换器。 In the preferred embodiment embodiment, the NIC driver to establish a redundant connection VI manner, by a pair of connecting the two switch fabrics 115a, b and the other by a further switch. 而且在首选实施例中,数据在该对的两个分支上交替发送,在交换器上均衡负载。 Further embodiment, the data transmitted alternately on the two branches of the preferred embodiment, load balancing on the switch. 替换地,冗余对也能以故障在线恢复方式使用。 Alternatively, redundancy for fault-line also can be used in recovery mode.

只要操作系统保持运行,由节点建立的所有连接对能够维持,模拟以太网连接的连接对的建立试图尽可能持久地类似于在网络接口卡之间实际地插入电缆。 Keep running as long as the operating system, established by all nodes connected to be able to maintain, simulate an Ethernet connection for connection establishment attempt as much as possible similar to actual lasting inserted cable between the network interface card. 若在节点的操作系统运行时节点的已定义的配置改变了,则在改变时刻可应用的冗余虚拟接口连接对被建立或舍弃。 If you have defined the node configuration changes in the node's operating system is running, you can apply at the time of changing the redundant virtual interface connection is established or discarded.

在控制节点端的千兆网驱动器逻辑325类似于上述那样地运行。 In the control node side Gigabit Ethernet driver logic 325 is similar to the operation described above.

虚拟LAN服务器虚拟LAN服务器逻辑335便于在底层NBMA网络上模拟以太网。 Virtual LAN Server Virtual LAN server logic 335 on the bottom layer facilitates NBMA Ethernet network simulation. 虚拟LAN服务器逻辑:2、管理相应虚拟LAN的成员;3、提供RVI映射及管理;4、对RVI的ARP处理及IP映射;5、提供广播及多播服务;6、便于桥接和路由到其他域;和7、管理服务群集。 Virtual LAN server logic: 2, manage the corresponding virtual member of the LAN; 3, providing RVI mapping and management; 4, RVI-ARP processing and IP mapping; 5, providing broadcast and multicast service; 6, to facilitate bridging and routing to other domain; and 7, the cluster management services.

1、虚拟LAN成员管理管理员使用管理应用程序135配置虚拟LAN。 1, a member of the virtual LAN management administrator using a management application 135 configuring virtual LAN. 在虚拟LAN上IP地址的分配和配置能以如“普通“子网的同样方式完成。 IP address allocation and configuration can be accomplished in the same manner as "normal" on the virtual subnet LAN. 使用的IP地址的选择取决于虚拟LAN上节点的外部可见性。 Selected depends on the IP address using the virtual node on the LAN external visibility. 若虚拟LAN不是全局可见(或在平台外不可见,或从因特网不可见),应使用专用IP地址。 If the virtual LAN is not globally visible (or invisible outside the platform, or are not visible from the Internet), you should use private IP addresses. 否则,IP地址必须从由提供因特网连接的因特网服务提供者(ISP)提出的范围配置。 Otherwise, IP addresses must provide the Internet service provider (ISP) range proposed by configuring the Internet connection from the. 通常,虚拟LAN IP地址的指定必须象正常LAN IP地址指定那样相同地处理。 Typically, virtual LAN IP address specified must be specified in the same manner as the processing as normal LAN IP address. 存储在控制节点120的本地磁盘上的配置文件定义在虚拟LAN中的IP地址。 Configuration files stored on the local disk 120 of the control node IP address is defined in a virtual LAN. 为了虚拟网络接口的目的,IP的别名只是建立在虚拟LAN服务器逻辑335上另外IP到RVI的映射。 For the purpose of the virtual network interface, the IP alias names to create additional IP RVI mapping on the virtual LAN server logic 335. 每个处理器在需要时能配置多个虚拟接口。 Each processor when needed can configure multiple virtual interfaces. 在建立和配置虚拟网络接口方面的主要限制是IP地址分配和配置。 The main limitation in the interface to build and configure a virtual network is an IP address allocation and configuration.

每个虚拟LAN具有服务器逻辑335的对应实例,它在控制节点120上和一系列在处理器节点105上执行的节点上执行,由管理员定义拓扑结构。 Each instance has a corresponding virtual LAN server logic 335, which performs control on the node 120 and a series of nodes executing on processor node 105, the topology defined by the administrator.

每个虚拟LAN服务器335被配置成只管理一个广播域,在给定的第二层广播域上能出现任何数目的第三层(IP)子网。 Each virtual LAN server 335 is configured to manage only one broadcast domain, it can be any number of third layer (IP) subnet on the second layer of a given broadcast domain. 响应管理员的命令配置和建立服务器335,以建立虚拟LAN。 Administrator response commands to configure and build server 335 to establish a virtual LAN.

当处理器106自举并配置其虚拟网络时,它通过专门的管理RVI连接到虚拟LAN服务器335。 When the bootstrap processor 106 and configure the virtual network, which is connected to the virtual LAN server 335 via a dedicated management RVI. 然后处理器获得它们的数据链路配置信息,如对它指定的虚拟MAC地址,虚拟LAN成员信息等。 The processor then obtain their data link configuration information such as its assigned virtual MAC address, a member of the virtual LAN information. 虚拟LAN服务器335将判断和确认,试图连接它的处理器正是服务器335要服务的虚拟LAN的成员。 Virtual LAN server 335 will determine and confirm that its members are trying to connect is the server processor 335 to service the virtual LAN. 若该处理器不是虚拟LAN的成员,到服务器的连接被拒绝。 If the processor is not a member of a virtual LAN, the connection to the server is denied. 若是成员,虚拟网络驱动器310用虚拟LAN服务器注册其IP地址。 If the members of the virtual network driver 310 to register its IP address with the virtual LAN server. (在配置驱动器310时由IP堆栈305提供IP地址)。 (IP address provided when configuring the drive 310 by the IP stack 305). 然后虚拟LAN服务器将该IP地址联系到注册到达的RVI。 Then contact the Virtual LAN server IP address registered to RVI arrival. 这就使虚拟LAN服务器能找到与特定IP地址相关的处理器。 This makes virtual LAN server processor can be found associated with a particular IP address. 此外,IP地址与处理器的联系能通过虚拟LAN管理接口135实现。 In addition, contact with the processor through the IP address of the virtual LAN management interface 135 to achieve. 后一种方法对于用专门的处理恰当地配置群集IP地址或IP地址是必要的。 The latter method is necessary for special processing appropriately configure the cluster IP address or IP address.

2、RVI映射及管理如上描述,某些实施例在数据链路层使用RVI连接节点,并形成控制连接。 2, RVI mapping and management as described above, some embodiments use RVI connection node of the data link layer, and forming a control connection. 某些连接作为控制节点自举和初始化的一部分被建立和指定。 Some self-connection is established and designated as part of the initialization and lift control node. 为上述的理由使用数据链路层。 Using the data link layer above reasons. 使用控制连接交换管理,配置及健康信息。 Using the control connection to exchange management, configuration and health information.

某些RVI连接在用于单播通讯的节点之间,如2121-2。 Some RVI is connected between node for unicast communication, such as 2121-2. 其它RVI连接是到虚拟LAN服务器逻辑335的,使得服务器能处理请求,如ARP通讯,广播等。 RVI is further connected to the virtual LAN server logic 335, so that the server can handle the request, such as ARP communications, broadcasting or the like. 为建立RVI,虚拟LAN服务器335通过调用千兆网交换管理器360(用交换网和千兆网NIC提供)建立和去除RVI。 For the establishment of RVI, virtual LAN server 335 360 (provided by the switching network and Gigabit Ethernet NIC) to establish and remove RVI Gigabit Ethernet switching by calling the manager. 交换管理器能在控制节点120执行,并协同千兆网驱动器建立RVI。 Switching manager to be executed in a control node 120, and establish a synergy Gigabit Ethernet driver RVI.

如上所述,关于处理器连接,作为对虚拟LAN服务器335注册的节点,虚拟LAN服务器对该节点建立并指定虚拟MAC地址。 As described above, connected to the processor as a virtual LAN server node 335 registered to establish a virtual LAN server and specifies the MAC address of the virtual node. 与此相关,虚拟LAN服务器逻辑维护反映对各个节点的拓扑结构和MAC分派的数据结构,然后虚拟LAN服务器逻辑对节点间的单播路径建立对应的RVI。 In this connection, the virtual LAN server logic maintains the data that reflects the topology structure of each node and the MAC assigned, then establishing a virtual LAN server logic RVI corresponding unicast path between nodes. 这些RVI随后被分配,并在节点自举过程中通知各节点。 The RVI is then assigned, and notifies each node in the node bootstrap process. 而且,RVI还与虚拟LN服务器处理ARP通讯期间的IP地址有关。 Furthermore, RVI also deal with IP addresses during ARP LN communication related to the virtual server. 若节点从拓扑结构中移去,RVI连接断开。 If the node is removed from the topology, RVI disconnected.

若在所建立的RVI连接一端的节点106重自举,该连接每一端的两个操作系统和RVI管理逻辑重新建立连接。 If one end of a connection node in the established RVI 106 weight bootstrap, the two operating systems and RVI connection management logic reestablish the connection at each end. 使用在运行中的处理节点上连接的软件不会觉察到连接本身发生的任何事情。 Software connected on use in the operation of processing nodes are not aware of anything that happened in the connection itself. 软件是否注意或关心在另一端的软件的重自举取决于它为什么使用连接,及在什么程度上该重自举端能从永久存储器重建其状态。 Attention or care whether the software at the other end of the bootstrap software re-use depends on why it is connected, and to what extent the heavy bootstrap end rebuild its state from persistent storage. 例如,任何通过传输控制协议(TCP)通讯的软件应注意到,在重自举时所有TCP会话被关闭。 For example, any of Transmission Control Protocol (TCP) communications software should be noted that all TCP session is closed when the re-bootstrap. 相反,网络文件系统(NFS)访问是无状态的,若重自举发生在允许的超时周期内,它不受重自举的影响。 In contrast, the Network File System (NFS) access is stateless, when the weight bootstrap occurs within the allowed time-out period, it is not the bootstrap heavy impact.

若一节点不能在任何时刻在直接的RVI上发送包,它总能尝试通过虚拟LAN服务器335发送包到目的地。 If a node can not send the packet directly on the RVI at any time, it always tries to send the packet to the destination 335 through the virtual LAN server. 因为虚拟LAN服务器335通过控制连接连接到虚拟LAN上的所有虚拟以太网驱动器310接口,虚拟服务器335也能用作最后手段的包转接机制。 Because the virtual LAN server 335 via a control connection connected to all the drives on the virtual Ethernet virtual LAN interfaces 310, 335 can also serve as a virtual server packet forwarding mechanism of last resort.

关于虚拟LAN服务器335的连接,某些实施例使用虚拟以太网驱动器310,它在算法上确定,为连接到其相关的虚拟LAN服务器335它应使用的RVI。 About Connection virtual LAN server 335, some embodiments use virtual Ethernet driver 310, it is determined algorithmically, RVI connected to its associated virtual LAN server 335 that it should use. 取决于实施例的算法可能需要考虑标识信息,如识别RVI的机柜号。 Depending on the embodiment of the algorithm may need to consider the identification information, such as the identification number RVI cabinet.

3、ARP处理和IP到RVI的映射如上解释,某些实施例的虚拟以太网驱动器310支持ARP。 3, ARP and IP mapping process explained above RVI, some virtual Ethernet driver 310 of the embodiment supports ARP. 在这些实施例中,使用ARP处理有助于在节点上建立在IP地址和RVI之间的映射,而RVI能用于在节点间传输包括IP包的单播通信。 In these embodiments, using the ARP processing helps to establish a mapping between the IP address and RVI on the node, and RVI can be used in a transmission packet including the IP unicast communication between nodes.

为此,虚拟以太网驱动器310经专用的RVI发送ARP包的请求及回答到虚拟LAN服务器335,虚拟LAN服务器335,和ARP服务器逻辑355通过加信息到包的标题而处理该包。 For this purpose, the virtual send an ARP request packet 310 via a dedicated Ethernet driver and RVI answer to the virtual LAN server 335, the virtual LAN server 335, and server logic 355 ARP packet header to the packet processed by the additional information. 如上解释,此信息便于识别源和目标,并识别可在节点之间使用的RVI。 As explained above, this information is easy to identify the source and destination, and identifies RVI can be used between nodes.

ARP服务器逻辑355接收ARP请求,处理TLV标题,并将请求广播到内部平台和合适的外部网络上的所有相关节点。 ARP server logic 355 receives the ARP request, the processing TLV header and the request is broadcast to all nodes on the inner platform and the appropriate external network. 除其它事情以外,服务器逻辑确定谁应接收根据请求导致的ARP回答。 Among other things, server logic to determine who should receive answers in accordance with a result of the ARP request. 例如,若源是群集的IP地址,回答应发送到群集负载均衡器,而不必是ARP请求的源。 For example, if the source IP address of the cluster, the cluster answers should be sent to the load balancer, but not necessarily the source of the ARP request. 服务器逻辑355通过将信息包括在ARP请求的TLV标题来指出那里,从而向ARP的目标回答。 Server logic 355 by the information included in the ARP request TLV title to point out there, so the answer to the ARP target. 服务器335通过将更多信息包括在附加的标题中来处理ARP包,并广播该包到有关域中的节点,例如,修改的标题能包括识别源机柜,处理节点号,RVI连接号,通道,虚拟接口号,和虚拟LAN名(其中某些只由服务器335知道)的信息。 More information by the server 335 included in the additional header to process ARP packets, and broadcasts the packet to the nodes of the relevant domain, e.g., a modified header identifying the source energy comprises a cabinet, processing node number, the RVI connection number, channel, information virtual interface number, name, and virtual LAN (some of which only 335 known by the server).

ARP的回答被服务器逻辑335接收,后者随后将回答中的MAC信息映射到对应RVI的有关信息。 ARP reply is received server logic 335, which then maps the MAC information about the answer information corresponding to the RVI. RVI的有关信息放在回答目标MAC条目中,并被送到适当的源节点(如可以是请求的发送者,但在如群集IP地址的某些情况,可以是其它节点)。 For information on the RVI to answer the destination MAC entries, and to the appropriate source node (e.g., sender of the request may, in some cases, such as the IP address of the cluster, may be other nodes).

4、广播和多播服务如上描述,通过在专用RVI上接收包来处理广播。 4, broadcast and multicast services described above, by receiving a broadcast packet to process on a dedicated RVI. 然后包被服务器335克隆,并单播到相关广播域中的所有虚拟接口310。 Then coated cloned server 335, and unicast related to the virtual broadcast domain to all interface 310.

对多播可使用同样的方法。 Multicast the same method can be used. 所有多播包将从虚拟LAN服务器反射出来。 All multicast packets from the virtual LAN servers reflected. 在某些另选的实施例中,虚拟LAN服务器象广播那样处理多播,并依赖在每个节点的IP过滤,过滤出不要的包。 In certain alternative embodiments, the virtual LAN server so as broadcast multicast processing, and each node relies on IP filtering, filter out unwanted packets.

在应用程序希望发送或接收多播地址时,它必须首先加入多播组。 When the application wants to send or receive a multicast address, it must first join a multicast group. 在处理器上的处理实现多播加入时,处理器虚拟网络驱动器310通过专用RVI发送加入请求到虚拟LAN服务器335。 Implemented on a processor processing the multicast join, the processor 310 via a virtual private network drive RVI sending join request 335 to the virtual LAN server. 然后如下面讨论,虚拟LAN服务器在接口上配置特定的多播MAC地址,并必要时通知LAN代理340。 Then, as discussed below, the virtual LAN server configuration specific multicast MAC address on the interface, and when the agent 340 notifies LAN necessary. 代理340必须保持在特定多播组上使用计数的跟踪,所以当没有处理器属于该多播组时多播地址被去除。 Agent 340 must keep track of the usage count on a particular multicast group, if it is not a processor belonging to the multicast address of the multicast group when being removed.

5、桥接和路由到其它域从系统100的图中,外部网络125能以两种方式运行,过滤的或非过滤的。 5, bridging and routing to other domains, from the external network 125 can run the system 100 in FIG two ways, filtered or non filtered. 在过滤方式中,使用整个系统的单个MAC地址于所有输出的包。 A single MAC address in the filtering mode, used to package all of the system output. 这就隐藏在虚拟LAN代理340背后的处理节点107的虚拟MAC地址,并使得系统看来象在网络125上的单个节点(或在网桥或代理背后的多个节点)。 This virtual LAN hidden behind the proxy processing node 340 of the virtual MAC address 107, and so the system appears as a single node on the network 125 (or multiple bridges or nodes in the back of the agent). 因为这没有对每个内部节点107暴露了单独的链路层信息,需要某些其它单独的标识符号以便适当地提交进入的包。 107 exposed because no separate link layer information for each internal node, need some other identification symbol in order to properly separate submitted incoming packets. 当以过滤方式运行时,使用每个进入包的目标IP地址来单独地识别预定的接收者,因为MAC地址上识别系统。 When operating in the mode when the filter, using the destination IP address to every incoming packet individually identify the intended recipient, because the MAC address of the identification system. 在非过滤方式中,节点107的虚拟MAC在系统之外是可见的,所以它们能用于直接进入的通信。 In the non-filtering mode, the nodes of the virtual MAC 107 is visible outside the system, so that they can be used directly to communications. 即过滤方式委托第三层交换而非过滤方式允许第二层交换。 That is entrusted third exchange filter, not allowing Filter Layer 2 switching. 过滤方式需要某些组件(在此情况是虚拟LAN代理340)实现在所有输出的包中用外部网络125的MAC地址替代节点的虚拟MAC地址。 Filter require certain components (in this case a virtual LAN agent 340) of virtual MAC addresses substitute node external network 125 in the packet with all outputs.

某些实施例支持连接到外部网络的虚拟LAN能力。 Certain embodiments support connected to an external network virtual LAN capability. 因此,虚拟LAN必须处理未在本地配置的IP地址。 Therefore, the virtual LAN must handle IP address is configured locally. 为此,一个实施例加上对外部广播域,那样连接的每个虚拟LAN受限制的极限。 For this purpose, one case coupled with an external broadcast domains, limit each virtual LAN connected as limited embodiments. 对虚拟LAN的内部节点,IP地址和子网的指定必须与外部域一致。 The internal nodes of the virtual LAN, and specify the IP address and subnet must be consistent with external domains.

虚拟LAN服务器335通过有效地起着链路层网桥的作用服务于外部连接,在网桥中在外部以太网驱动器345和内部处理器之间传送包,并实现非IP处理。 Virtual LAN server 335 via the link layer functions effectively service the external connection bridge, the bridge in the packet transmitted between the external and internal Ethernet driver 345 processor, and to achieve a non-IP processing. 但是不像数据链路层的网桥,服务器不能总是依靠从外部网络到内部节点的完全不同层的两个地址,不同的是连接可以使用第三层(IP)信息作出桥接决策。 However, unlike the data link layer of the bridge, the server can not always rely entirely different from the external network to the internal nodes of layer two addresses, except that the connection may use a third layer (IP) information to make decisions bridge. 为了这样做,外部连接软件从进入的包抽取IP地址,且它使用此信息识别正确的节点106,使包能传送到那个节点。 To do so, the external connection software extracted from incoming packets IP address, and it uses this information to identify the proper node 106, so that the packet can be transmitted to that node.

具有连接的外部广播的虚拟LAN服务器335必须截取并处理来往外部域的包,使得在广播域中外部节点具有子网的一致的视图。 Virtual LAN server 335 has an external broadcast connection must intercept and process the packet to and from an external domain, such that a consistent view of the external node with subnet broadcast domain.

当具有连接的外部广播域的虚拟LAN服务器335从外部节点接收ARP请求时,它将转接该请求到所有内部节点。 When virtual LAN server 335 has an external connection to receive broadcast domain ARP request from an external node, it will transfer the request to all internal nodes. 随后正确的节点将组成回答并将回答通过虚拟LAN服务器335送回到请求者。 Followed by the correct node will form the answer and the answer sent back to the requester through the virtual LAN server 335. 虚拟LAN服务器与虚拟LAN代理340合作,使得该代理在输出的请求上处理任何必要的MAC地址翻译。 Virtual LAN with a virtual LAN server agent 340 cooperate so that any necessary processing agent MAC address translation on the request output. 从外部源来的所有ARP回答及ARP广告将直接转接到目标节点。 Coming from an external source and answer all ARP ARP advertising will be directly transferred to the destination node.

虚拟以太网接口310将在控制连接RVI上发送所有带有外部目标的单播包到虚拟LAN服务器335。 Virtual Ethernet interface 310 sends all targets with external unicast packet to the virtual LAN server 335 on the control connection RVI. (外部目标可由驱动器借助MAC地址格式辨别)。 (External target discrimination by the drive means of the MAC address format). 然后虚拟LAN服务器传送包到外部网络125。 Then the virtual LAN server 125 transmits the packet to the external network.

若虚拟LAN服务器335从内部节点接收广播或多播包,除了转接该包到所有内部虚拟LAN成员外,它将包转接到外部网络。 If the Virtual LAN server 335 receives a broadcast or multicast packet from an internal node, in addition to forwarding the packet to all the members of the internal virtual LAN, the packet will be transferred to an external network. 若虚拟LAN服务器335从外部源接收广播或多播包,它转接该包到所有连接的内部节点。 If the Virtual LAN server 335 receives a broadcast or multicast packet from an external source, which transfer the packet to all internal nodes.

在某些实施例中,通过采用IP路由器和防火墙,使用在互连的物理的LAN中类似的机制,实现互连的虚拟LAN。 In certain embodiments, through the use of IP routers and firewalls, using similar physical interconnect mechanism in a LAN, interconnected virtual LAN. 在双方LAN上配置一个处理器,在该处理器上的Linux核心必须启用路由(可能是IP的伪装)。 A processor configured on both the LAN, Linux kernel on the processor must be enabled routing (IP could be disguised). 即使两个节点位于同一平台上,正常的IP子网和路由语义要永久维护。 Even if the two nodes on the same platform, the normal IP subnet and routing semantics to be permanently maintained.

一个处理器应配置成在两个外部子网之间、在外部和内部子网之间,和在两个内部子网之间的路由器。 A processor configured to be between two external subnet, between the external and internal subnets, routers, and between the two internal subnets. 由于内部网络的点对点拓扑结构,当内部节点通过路由器发送包时不存在问题。 Since the point of the topology of the internal network, the internal node when the packet is transmitted via a router is not a problem. 发送者直接发送到路由器(即用路由逻辑配置的处理器)而没有虚拟LAN服务器的干涉(即如上讨论典型的处理器到处理器的通讯)。 Sender sent directly to the router (i.e. the routing processor configured with logic) without interfering virtual LAN server (i.e., as discussed above, a typical processor to the communication processor).

当外部节点发送包到内部路由器,且外部网络以过滤方式运行时,进入包的目标MAC地址将是平台100的MAC地址。 When the external node sends the packet to the internal router, and the external network when operating in a filtering mode, the destination MAC address of the incoming packet will be the MAC address 100 of the platform. 因此MAC地址不能被用于唯一地识别包的目标节点。 Thus MAC address can not be used to uniquely identify the destination node of the packet. 对于其目标是虚拟LAN上的内部节点的包,在IP标题中的目标IP地址被用于将包引导到适当的目标节点。 The goal is for the internal node packet on the virtual LAN, destination IP address in the IP header is used for the packets to appropriate destination node. 但是,因为路由器不是最终目标,在IP标题中的目标IP地址是最终目标的地址,而不是下一个转发段(内部路由器)的地址,因此在进入的包中没有什么能用于引导它到正确的内部节点。 However, because the router is not the final goal, the destination IP address in the IP header is the address of the ultimate goal, rather than a forwarding address of the next segment (internal router), so nothing can be used to guide it into the correct package the internal nodes. 为处理此情况,一个实施例加了一个限制,在虚拟LAN中最多只有一个路由器暴露在外面网络中。 To handle this situation, one embodiment of a restriction on the embodiment, a virtual LAN, a router at most exposed to the outside network. 此路由器用虚拟LAN服务器335注册成默认的目标,使得无有效目标的进入包将向此默认节点。 This virtual LAN router server 335 registered as the default destination, such that no valid object will enter this default packet node.

当外部节点发送包到内部路由器,且外部网络125以非过滤方式运行,输入包的目标MAC地址将是内部目标节点的虚拟MAC地址。 When the external node sends the packet to the internal router, and the external network 125 operating in a non-filtered mode, the input packet destination MAC address is a virtual MAC address of the internal destination node. 然后LAN服务器335将使用此虚拟MAC地址直接发送包到目标内部节点。 Then the server LAN 335 using the virtual MAC address to send packets directly target internal node. 在此情况,任何数目的内部节点都能起到路由器的作用,因为输入包的MAC地址将唯一地识别该目标地址。 In this case, any number of internal nodes are able to function as a router, because the input packet to the MAC address uniquely identifying the destination address.

若配置需要在一个子网中有多个路由器,一个路由器能选作暴露的路由器。 If the configuration requires a plurality of routers in a subnet, the router can be selected as a router exposed. 在必要时此路由器能转而路由到其它路由器。 In turn, if necessary, this router can route to other routers.

在本发明的某些实施例中,基于无状态(即,每个IP包而非每个TCP连接)使路由器成为一个群集服务并且均衡负载或故障在线恢复来提供路由器冗余。 In certain embodiments of the present invention, based on a stateless (i.e., not every IP packet for each TCP connection) so that the router becomes a cluster service line and load balancing or fault recovery provides router redundancy.

本发明的某些实施例通过提供交换语义支持混合的模式的功能,其中给定的端口能被设计成混合端口,使得所有通过交换机的通信量在混合端口上重复。 Certain embodiments of the present invention by providing a supported mixed mode switch semantic features, wherein a given port of the port can be designed as a hybrid, such that repeated on all traffic mixed port via the switch. 在虚拟LAN服务器上在管理上应指定允许以混合模式监听的节点。 On a virtual LAN server on the management node should be specified in order to allow mixed-mode listening.

当虚拟以太网接口310进入混合接收模式,它将在管理RVI上发送报文到虚拟LAN服务器335。 When mixed into the virtual Ethernet interface 310 receiving mode, it sends the packet to the virtual LAN on the management server 335 RVI. 此报文包含所有有关进入混合模式的虚拟以太网接口310的信息。 This message contains all information about entering mixed-mode virtual Ethernet interface 310. 当虚拟LAN服务器从一个节点接收混合模式的报文,它校验其配置信息,判断是否允许该节点混合地监听。 When the server receives the virtual LAN packets from a mixed mode node, checking that its configuration information, the node determines whether to allow mixed listens. 若否,虚拟LAN服务器将丢下此混合模式的报文不再处理。 If not, the virtual LAN server will leave this mixed-mode messages are no longer processed. 若节点允许进入混合模式,虚拟LAN服务器将向虚拟LAN上的所有其节点广播该混合模式报文。 If the node allowed to enter hybrid mode, the virtual server to the virtual LAN on all nodes in the mixed-mode broadcast packets on the LAN. 虚拟LAN服务器也将标记该节点为混合的,使得能向它转送输入的外部包的拷贝。 Virtual LAN Server will also mark the node is mixed, so that the copy can be transferred to the input of its outer cladding. 当混合的监听者检测到在其RVI配置中的任何改变,它将发送混合模式的报文到虚拟LAN,更新在相关广播域中所有其它节点的状态。 When the mixing of the listener to detect any change in its RVI configuration, it sends a mixed mode of the LAN packet to the virtual, update the status of all other nodes associated broadcast domain. 这将更新任何进入或离开虚拟LAN的节点。 This will update any entering or leaving the virtual LAN nodes. 当虚拟以太网接口310离开混合方式,它将向虚拟LAN服务器发送一报文,通知它接口正离开混合模式。 When the virtual Ethernet interface 310 leaving the mixing mode, it will send a packet to the virtual LAN server, it notifies the interface is leaving the mixed mode. 然后虚拟LAN服务器发送此报文到该虚拟LAN上的所有其它节点。 Virtual LAN server then transmits the packet to all other nodes on the virtual LAN. 当任何内部虚拟接口是混合方式监听者时,混合设定考虑以混合模式放置一个外部连接。 When any internal virtual listener interfaces are mixed manner, mixing for external connection to consider placing a mixed mode. 这使平台外部的通信量(但在同一虚拟LAN上)为混合的监听者可用。 This allows the external traffic platform (but on the same virtual LAN) available mixed listener.

6、管理服务群集服务群集是在一个或多个IP地址(或主机名称)上可得到的一组服务。 6, service management cluster service cluster is a group of one or more services on the IP address (or host name) available. 这些服务的例子是HTTP,FTP,远程登录,NFS等。 Examples of these services are HTTP, FTP, Telnet, NFS, etc. IP地址和端口号对表示由群集向包括外部网络125上的客户提供的特殊服务类型(虽然不是服务实例)。 IP address and port number indicating the particular type of service provided to the client 125 include an external network by the cluster (although not the service instance).

图5示出某些实施例如何通过群集IP地址将虚拟的服务群集505表示成对因特网或其它外部网络125的单个虚拟主机。 Figure 5 illustrates how some embodiments of the cluster IP address 505 represents the virtual cluster service paired Internet or other external network 125 is a single virtual hosts. 群集505的所有服务通过单个IP地址,通过在该IP地址的不同端口编址,在图5的例中,服务B是负载均衡服务。 All the service cluster 505 through a single IP address, by addressing the different ports of the IP address, in the embodiment of FIG. 5, the service B is a load balancing service.

参考图3B,虚拟群集由与虚拟LAN服务器335合作的虚拟群集代理(VCP)逻辑360的合成所支持。 3B, the virtual cluster is supported by the virtual cluster agent cooperation with the virtual LAN server 335 Synthesis of (VCP) logic 360. 简而言之,VCP360负责对每个配置的虚拟IP地址处理输入连接,端口过滤器,和真实服务连接的分配。 In short, VCP360 responsible for allocating input connection, port filter, and real service connection to the virtual IP address of each configuration process. 对每个配置的群集IP地址存在一个VCP。 There is a VCP cluster IP address for each configuration.

当一个包到达虚拟IP地址,虚拟LAN代理逻辑将发送该包到VCP360进行处理。 When a packet arrives at the virtual IP address, virtual LAN proxy logic to send the packet to be processed VCP360. 然后VCP根据包的内容,它的内部连接状态缓冲器,应用到输入通信的负载均衡算法,和构成的服务的可用性决定将包发送到何处。 Then the VCP packet according to the content of its internal connection state of the buffer, load balancing algorithm applied to the incoming communication, and determines service availability configuration where to send the packet. VCP将根据目标IP地址和TCP或UDP端口号转接输入的包。 VCP according to destination IP address and TCP or UDP port number of the packet input adapter. 此外,它只分配为VCP已知的端口号(或对存在的TCP连接)指定的包。 Furthermore, it is only known VCP assigned port number (or connection to existing TCP) specified package. 正是这些端口的配置和端口号对一个或多个处理器的映射建立了虚拟的群集,并使特定的服务实例在群集中可得到。 It is these configuration and number of the port on a map or more processors to establish a virtual cluster, and specific examples of centralized services available in the cluster. 若从多个应用服务器构造同样服务的多个实例,VCP能在服务实例之间均衡负载。 If multiple instances of the application server from the plurality of constructed the same service, VCP can load balance between service instances.

VCP360维护在群集的IP地址上存在的所有活动的连接的缓冲器。 VCP360 maintain all activities occurring on the IP address of the cluster is connected to a buffer. 任何作出的负载均衡决策只有当在客户和服务之间建立新的连接时作出。 Any load balancing decisions to be made only when a new connection is established between the client and service. 一旦建立了连接,VCP将使用在输入包的标题中的源和目标信息保证在TCP流中的所有包路由到提供服务的同一处理器106。 Once the connection is established, VCP uses source and destination information in the header of the input packet routes to ensure that all packets in the same TCP flow to the processor 106 to provide services. 在没有能力确定客户会话(如HTTP会话)的情况,实际的连接/负载均衡映射缓冲根据客户地址路由数据包,使得从同一客户的随后连接进到同一处理器(使客户会话持续或“粘住”)。 Without the ability to determine client sessions (e.g. HTTP sessions), the actual connection / load balancer mapping the address buffer according to the customer routing packets, so that the connection from the same client then proceeds to the same processor (enable client session duration or "sticking "). 在服务端口号基础上会话持续应是可选的,因为只有某些类型的服务需要会话持续。 On the basis of port services on the session duration should be optional, because only certain types of services that require session persistence.

对ARP请求的回答和ARP回答的路由由VCP处理。 Answer ARP requests and ARP reply is processed by the routing VCP. 在处理器发送任何ARP包时,它通知虚拟以太网驱动器310发出。 When the processor sending any ARP packet, it notifies the virtual Ethernet driver 310 issued. 然后包被送到虚拟LAN服务器335进行正常的ARP处理。 Then the packet is sent to the virtual LAN server 335 normal ARP processing. 通常虚拟LAN服务器广播此包,但保证不会广播到群集的任何成员(不只是发送者)。 Usually virtual LAN server broadcasts this package, but will not be broadcast to ensure that any member of the cluster (not just the sender). 它也将信息放在包的标题TLV,向ARP目标指出,通过虚拟LAN服务器,尤其是通过负载均衡器只能达到ARP源。 It will also be information in the packet header TLV, the ARP target noted by the virtual LAN server, especially through the load balancer can only reach the ARP source. 无论是内部或外部的ARP目标正常地处理ARP请求,并通过虚拟LAN服务器将回答送回。 Either internal or external target ARP ARP request is processed normally, and virtual LAN server will answer back. 因为ARP的源是群集IP地址,虚拟LAN服务器不能确定哪个处理器发出原始的请求。 Because the source IP address of the ARP is a cluster, the virtual LAN server can not determine which processor issued the original request. 因此虚拟LAN服务器将回答送到每个群集成员,使它们能正确地处理它。 Therefore, the virtual LAN server will respond to each cluster member, so that they can properly handle it. 当ARP包被带着群集IP地址作为目标的源发送时,虚拟LAN服务器发送请求到每个群集成员。 When the source cluster sends an ARP packet with the IP address as the destination, the virtual LAN server sends a request to each cluster member. 每个群集成员接收该ARP请求并正常地处理它。 Each cluster member receives the ARP request and process it properly. 然后它们组成ARP回答并经过虚拟LAN服务器发回给源。 Then they form a virtual LAN ARP reply and after the server sends back to the source. 当虚拟LAN服务器从群集成员接收任何ARP回答时,它丢弃该回答,但虚拟LAN服务器将组成并发送ARP回答给ARP源。 When virtual LAN server receives the ARP reply from any of the cluster members, it discards the reply, but the composition of the virtual LAN and the server to send an ARP reply ARP source. 因此,虚拟LAN服务器将响应群集IP地址的所有ARP。 Thus, the virtual LAN server will respond to all ARP cluster IP address. ARP回答将包含为了ARP源将所有对该群集IP地址的包发送到VCP所必需的信息。 ARP reply containing information for ARP source VCP necessary to transmit all the packets of the IP address of the cluster. 对外部ARP源,这简单地是带着作为源硬件地址的外部MAC地址的ARP回答。 An external source of ARP, with ARP is that simply as an external source address of the MAC hardware address is answered. 对内部ARP源这是必要的信息,告诉源发送对群集地址的包到虚拟LAN管理RVI,而不是通过直接连接的RVI。 This is the source of internal ARP necessary information, tell the source of the package sent to the cluster address to the virtual LAN management RVI, rather than through a direct connection of RVI. 任何接收到的无理由的ARP包转送到所有群集成员。 Any gratuitous ARP packets received forwarded to all cluster members. 由群集成员发送的任何无理由的ARP包将被正常地发送。 Any gratuitous ARP packets sent by the cluster members will be sent normally.

虚拟LAN代理虚拟LAN代理340完成在所有具有的外部物理网络125的虚拟接口的处理器中物理的网络资源的基本协调。 Agent Virtual LAN Virtual LAN agent 340 to complete the basic physical coordination processor virtual network resources in physical network interfaces with all external of 125. 它将虚拟LAN服务器335桥接到外部网络125。 It will bridge the virtual LAN server 335 to an external network 125. 当外部网络125以过滤方式运行时,虚拟LAN代理340将内部虚拟MAC地址从每个节点转换成对系统100指定的单个外部MAC。 When the external network 125 to filter operation mode, the virtual LAN Agent 340 inside the single external virtual MAC MAC address specified in the paired system 100 converts each node. 当外部网络125以非过滤方式运行时,不需要那样的转换。 When the external network 125 operating in a non-filtering mode, the conversion does not require that. 虚拟LAN代理340还完成IEEE802.1Q虚拟LAN ID标签信息的插入和去除,并根据它们的VLAN Id分离包。 Virtual LAN agent 340 also complete the insertion and removal IEEE802.1Q virtual LAN ID tag information, and separated according to their packet VLAN Id. 它还使对物理的以太网接口129的访问顺序化,并协调在物理网络上如多播地址那样的MAC地址的分配及去除。 It also allows access to the physical order of the Ethernet interface 129, and to coordinate distribution and removal, such as the multicast address as a MAC address on the physical network.

当外部网络125以过滤方式运行,且虚拟LAN代理340从虚拟LAN服务器335接收输出的包(ARP或其它)时,它用物理的以太网设备129的MAC地址代替内部格式MAC地址,作为源MAC地址。 When the external network 125 to filter operation mode, and the virtual LAN packet 340 from the virtual LAN proxy server 335 receives the output (ARP or other), it instead of the internal format of the MAC address with the MAC address of the Ethernet physical device 129, as the source MAC address. 当外部网络125以非过滤方式运行时,不需要那样替代。 When the external network 125 in a non-operating mode when the filter is not required as an alternative.

当虚拟LAN代理340接收输入的ARP包,它传送该包到虚拟LAN服务器335,后者处理该包并将该包转接到正确的目标。 When virtual LAN ARP proxy 340 receives an input packet, it transmits the packet to the virtual LAN server 335, which processes the packet and the packet forwarded to the correct destination. 若ARP包是广播包,则包被转接到虚拟LAN上的所有内部节点。 If the ARP packet is a broadcast packet, the packet is forwarded to all internal nodes on the virtual LAN. 若该包为单播包,则该包仅被发送到目标节点。 If the packet is a unicast packet, then the packet is sent only to the target node. 当外部网络125以过滤方式运行时,目标节点由ARP包中的IP地址确定,或由ARP包的以太网标题中的MAC地址确定(不是ARP包的MAC地址)。 When the external network 125 to filter operation mode, the target node is determined by the IP address in the ARP packet, or determined (the MAC address is not an ARP packet) from the Ethernet MAC address in the packet header ARP.

物理LAN驱动器在某些实施例中,到外部网络125的连接是通过连接到控制节点的千兆位或100/10baseT以太网链路。 Physical LAN driver In certain embodiments, the connection to the external network 125 via the control node connected to the gigabit or 100 / 10baseT Ethernet link. 物理的LAN驱动器345负责与那样链路的接口。 The physical LAN driver interface 345 that is responsible for the link. 在接口上发送的包以正常方式排队等待设备,包括将包收入报路缓冲器。 Packet transmitted on the interface in the normal manner queuing device comprising the packet buffer, which reported revenue. 用于排队包的队列是由协议堆栈使用的队列,将包排队等待设备的发送例行程序。 Queue is a queue for queuing packets used by the protocol stack, the packet transmission device waiting routine. 对输入的包,包含该包的报路缓冲器循环进行,而包数据不需拷贝(虽然对多播操作在需要时克隆)。 Packet input buffer loop path message containing the package is, but without copying the data packet (multicast although clone operation needs). 在这些实施例中,在控制节点中能使用通常的Linux网络设备驱动程序而不必修改。 In these embodiments, the control node can be used in conventional Linux network device driver without modification. 这便于增加新的设备到平台中,而不需要附加的设备驱动程序的工作。 This makes it easy to add new devices to the platform, without the need for additional work device driver.

物理的网络接口345只与虚拟LAN代理340通讯。 Physical network interface 345 and the communication agent 340 of virtual LAN. 这就避免控制节点以任何方式使用外部连接,那样会干扰虚拟LAN的操作,从而改善了用户数据的安全性及隔离,即管理员也不能“嗅探”任何用户的包。 This avoids the use of an external control nodes connected in any way that would interfere with the operation of the virtual LAN, thereby improving safety and isolation of the user data, i.e., the administrator can not "sniffing" of any user of the package.

负载均衡及故障在线恢复在某些实施例,另外可以使用到外部网络125的冗余连接,使到外部网络125的两个冗余接口之间的包传输负载均衡。 Load balancing and fault recovery line In some embodiments, may additionally using redundant connections to the external network 125, external network Shidao two redundant packet transmission, load balancing between the interface 125. 另外的实施例通过在交替的控制节点上配置每个虚拟网络接口来负载均衡,使得虚拟接口在两个控制节点之间均匀分配。 Further embodiments by configuring each virtual network interface on alternate control node to load balancing, such that the virtual interface evenly distributed between the two control nodes. 另外实施例通过一个控制节点发送并通过另一个接收。 Further embodiments of the other by sending and receiving via a control node.

在过滤方式中,存在外部可见的MAC地址,对一组虚拟网络接口,外部节点向它发送包。 In the filter, the presence of externally visible MAC address, set to a virtual network interface, it sends the packet to the external node. 若那个适配器停止,则不仅该虚拟网络接口必须故障在线恢复到另外控制节点,而且MAC地址也要故障在线恢复,使得外部节点能继续发送包到已经在ARP高速缓冲中的MAC地址。 If the adapter is stopped, not only the virtual network interface have line fault recovery control node to another, and also the MAC address line failure recovery, such that the external node to continue to send the packet to the MAC address is already in the ARP cache. 在本发明的一个实施例中,当故障的控制节点恢复时,单个MAC地址被处理,且该MAC地址不必须在恢复时重新被映射。 In one embodiment of the present invention, when the control node failure recovery, a single MAC address to be processed, and the MAC address does not have to be re-mapped in the recovery.

在本发明的另外实施例中,借助允许在两个控制节点发送但只通过一个接收来实现负载均衡。 In a further embodiment of the present invention, the control means allows two nodes to transmit but to achieve load balancing by only one receiver. 故障在线恢复的情况为发送并通过同一控制节点接收。 Case of a fault recovery line to transmit and receive the same control node. 恢复情况通过恢复的控制节点发送,因为那不需要任何MAC处理。 Recovery by transmitting the control node recovery, because it does not require any MAC processing.

进行接收的控制节点具有用于过滤的IP信息和用于多播MAC配置的多播地址信息。 Having a control node receiving IP for filtering information and for configuration of multicast MAC multicast address information. 此信息对处理输入包是需要的且在接收控制节点故障时此信息应能故障在线恢复。 This information is needed to process the input packet and this information can be restored upon receiving the fault line control node failure. 若发送控制节点故障,虚拟网络驱动器只需要将发送输出包到接收控制节点。 When the transmission control node fails, the virtual network driver only needs to send the packet to the output of the receiving control node. 除了辨别发送控制节点已故障之外不需要专门的故障在线恢复处理。 In addition to identifying the transmitter has been outside the control node failure does not require a special recovery process online. 若故障的控制节点恢复,虚拟网络驱动器能重新开始发送输出包到恢复的控制节点而不必任何时加的专门恢复处理。 If the control node failure recovery, the virtual network drivers can restart sending the packet to the control node outputs the restored without any special extra time restoration process. 若接收控制节点故障,则发送控制节点必须承担接收接口的作用。 When receiving a control node failure, the sending node must assume the role of receiving the control interface. 为此,必须在其物理接口配置所有MAC地址,使包能够接收。 For this purpose, it must be in all the physical interface MAC address, so that packets can be received. 另选地,两个控制节点能具有在它们的接口上配置的同一MAC地址,但在控制节点准备接收包之前,在以太网设备上能由设备驱动程序在物理上禁止接收。 Alternatively, two control nodes can be configured with the same MAC address on their interfaces, but in preparation before the control node receives the packet, can be prohibited from receiving device on the Ethernet by a device driver physically. 这样,故障在线恢复将简单地启用在该设备上的接收。 In this way, the fault line recovery will simply be enabled on the receiving device.

因为当任何处理器加入多播组时,接口必须用多播MAC地址配置,多播信息必须在控制节点之间共享,使得故障在线恢复对处理器透明。 Because any processor when a multicast group is added, the interface must be configured with the multicast MAC address, multicast control information must be shared between the nodes, so that the transparent line fault recovery processor. 因为虚拟网络驱动器无论如何必须保持对多播组成员的跟踪,此信息在需要时总能经过虚拟LAN服务器为LAN代理可用。 Because the virtual network drive in any case must keep track of multicast group members, this information can always be through a virtual LAN server LAN agent is available when needed. 因此,接收故障在线恢复将导致多播组成员从虚拟网络驱动器被查询以便重建本地多播组成员表。 Therefore, the receiving line fault recovery will lead to a multicast group members queried from a virtual network drive in order to rebuild the local multicast group membership table. 此操作是低开销的,且除了在故障在线恢复及恢复期间不需要专门的处理,并在各控制节点之间不需要任何专门的数据复制。 This operation is a low cost, and in addition does not require line during the failure recovery processing and recovery specialized and does not require any special data replication between the control node. 当接收已经处于故障在线恢复而且故障的控制节点恢复时,只有发送将被移到已恢复的控制节点。 When the receiver is already in a failed online recovery and control node failure recovery, only the sender will be moved to the control node has been restored. 因此在虚拟网络接口上恢复的算法总是将发送移到已恢复的控制节点,让接收处理留在原处。 Therefore, the recovery on the virtual network interface algorithm always sends move control node has been restored to enable the receiver to remain in situ process.

虚拟服务群集也能使用负载均衡及故障在线恢复。 Virtual Cluster service can also use load balancing and fault-line recovery.

多播平台某些实施例允许机柜互相连接以形成更大的平台。 Multicast internet some embodiments allow the cabinet interconnected to form a larger platform. 每个机柜至少具有一个控制节点用于机柜间的连接。 Each enclosure having at least one control node for the connection between the cabinet. 每个控制节点包括处理本地连接和通信的虚拟LAN服务器335。 Each control node 335 comprises a virtual LAN server process local connection and communication. 一个服务器配置成主服务器,如位于带外部的虚拟LAN连接的节点的那个服务器。 A server configured as the master server, such as in the virtual LAN server node with external connections. 另一虚拟LAN服务器起着代理服务器或从属服务器的作用,使这些机柜的本地处理器能参与其中。 Another virtual LAN server functions as a proxy server or a slave server, these cabinets can participate in the local processor. 主服务器维护所有虚拟LAN状态和控制,而代理服务器则在处理器和主服务器之间转接数据包。 The primary server maintains all the status and control virtual LAN, and proxy forwarding data packets between the processor and the main server.

每个虚拟LAN服务器代理维护对每个主虚拟LAN服务器的RVI。 Each virtual LAN server agents maintain virtual LAN server for each primary RVI. 每个本地处理器连接到虚拟LAN服务器的代理服务器,好象它就是主服务器。 Each local processor is connected to the virtual LAN server, a proxy server, if it is the master server. 当处理器连接并注册IP和MAC地址时,代理服务器将用主服务器注册该IP和MAC地址。 When the processor is connected and a MAC address and IP registration, the proxy server will register the MAC address and IP master server. 这引起主服务器将地址从代理服务器联系到RVI。 This causes the primary server to contact an address from the proxy server to RVI. 因此,主服务器包含对所有内部节点的RVI连接,但代理服务器只包含对同一机柜的节点的连接。 Thus, the main server includes a connection to all internal nodes RVI, but only contains the proxy node connected to the same cabinet.

当在多机柜虚拟LAN中任何处的处理器发送任何包到其虚拟LAN服务器时,包将被转换到主服务器进行处理。 When the LAN at any processor to send any packet whose virtual LAN of virtual servers in a multi-rack, the packet will be converted to the primary server for processing. 然后主服务器对该包作正常处理。 The master server for normal processing of the packet. 主服务器转接该包到代理服务器,对多播和广播这是必要的。 The primary server to transfer the package to the proxy server, multicast, and broadcast it is necessary. 主服务器还根据单播包的目标IP地址和在代理服务器上的注册IP地址转接单播包。 Master also redirect unicast packet Unicast packets according to destination IP address and the IP address registered on the proxy server. 注意,在主服务器上,代理服务器连接看上去很象带有许多配置的IP地址的节点。 Note that, in the master server, the proxy server is connected with a node looks like many IP addresses configured.

网络管理逻辑在处理节点上没有运行如自举或核心排错的操作系统的期间,节点的顺序的控制台通信量和自举映象请求被位于处理节点的核心排错软件或BIOS的交换机驱动器码路由,以便管理在控制节点(未示出)运行的软件。 Network management logic is not running on a processing node such as the console traffic bootstrap procedure or during troubleshooting of the kernel of the operating system, and the bootstrap node map request core BIOS software debugger or the switch drive is positioned in the processing node routing code, to the control node management software (not shown) is running. 由那里能再次访问控制占通信量,或者从高速外部网125,或者通过控制节点的管理端口。 From there access control again accounted for traffic, or a high-speed external network 125 from, or by a control node management port. 从控制节点的本地盘或从外部SAN 130之外的部分能满足自举映象请求。 From the local control node or from the disc than the outside portion SAN 130 bootstrap image to meet the request. 在对处理节点能做任何事情之前控制节点120最好自举并正常运行。 Prior to processing node can do anything better control node 120 and the bootstrap operating normally. 控制节点本身从其管理端口自举或排错。 Management control node itself from its port bootstrap or troubleshooting.

某些客户希望通过将他们的管理端在需要时插入现场计算机来将控制器的自举及排错限止到只有本地访问。 Some customers want to manage by their end into the computer scene when needed to the bootstrap controller troubleshooting limiting, and that only local access. 其他人可以选择通过为了管理的目的建立与因特网适当地隔离的安全网络段来允许远程自举和排错,他们的管理端口插入到该网段。 Others may choose to allow remote bootstrapping and troubleshooting by management for the purpose of establishing the Internet properly isolated secure network segment, their management port inserted into the network segment. 一旦控制器被自举并正常运行,若管理员允许,对它及平台其余部分的其它管理功能可以从高速外部网125以及管理端口访问。 Once the bootstrap and the controller is operating normally, if permitted by the administrator, and other management functions to its rest platform can access and manage the high-speed external network port 125.

来往于每个处理节点105的顺序的控制台通信量在交换结构115上由操作系统核心驱动器发送到在控制节点120运行的管理软件。 Each processing node and from the console traffic sequence 105 transmitted by the operating system kernel driver 120 to the control node management software running on the switch structure 115. 从那里,任何节点的控制台通信量能从正常的高速外部网络访问或通过双方控制节点的管理端口访问。 From there, the console traffic from any node of the normal high-speed external network access control management port or through mutual access node.

存储结构某些实施例遵循存储的SCSI模式。 Storage structure of some embodiments follows the SCSI storage mode. 每个虚拟PAN具有其自己的虚拟化I/O空间并在那个空间中发出SCSI命令及状态。 Virtualized I each has its own virtual PAN / O space and issues SCSI commands and status in that space. 在控制节点的逻辑必要时翻译或转换从PAN来的地址及命令,并接着将它们发送到服务这些命令的SAN 130。 Translation of the logical control node or from the need to address and PAN commands, and then sends them to the service of these commands SAN 130. 从SAN的角度,客户是平台100,而发出命令的实际PAN被隐藏和匿名。 SAN from the perspective of the client platform is 100, the actual command issued PAN is hidden and anonymous. 因为SAN空间是虚拟化的,在平台100上运行的PAN能具有从设备号1开始的设备编号,且第二个PAN也能具有设备号1。 Because space is SAN virtualization, running on the platform 100 PAN can have a number for the start device No. 1, and the second PAN also has a device number 1. 此外每个设备号1对应于SAN存储的不同的唯一部分。 In addition, each device 1 correspond to a different part of the unique SAN storage.

在首选实施例中,管理员能建立虚拟存储。 In the preferred embodiment example, the administrator can create virtual storage. 每个PAN具有它自己的海量存储的独立联系。 Each PAN has its own independent connection of mass storage. 因此如下面解释,第一PAN能具有到SAN中第一位置的给定的设备/LUN映射地址,而第二PAN具有到SAN第二个不同位置的同样的给定的设备/LUN映射。 Therefore, as explained below, you can have the first PAN to the SAN equipment in a first given position / LUN address mapping, similar to the second PAN having a given apparatus different positions of a second SAN / LUN mapping. 每个处理器将设备/LUN地址映射到主要和次要设备号,例如用于识别盘和分区。 Each processor maps the device / LUN address to the major and minor device number, for example for identifying a disc and partitions. 虽然主要和次要设备号被PAN和PAN中的处理器理解成物理地址,实际上它们被平台处理成对由SAN提供的海量存储的虚拟地址。 While the major and minor device number and PAN PAN be understood to a physical address in a processor, the virtual address is actually stored in the mass treated in pairs which are provided by the internet SAN. 即每个处理器的主要和次要设备号被映射到对应的SAN位置。 I.e., each processor major and minor device number is mapped to a corresponding position of the SAN.

图6示出用于实现某些实施例的存储结构的软件组件。 Figure 6 illustrates the software components used to implement some embodiments of the memory structure. 通常,在控制节点120执行的配置组件605与外部SAN 130通讯。 Typically, 130 in communication with an external configuration component 605 SAN 120 performed by the control node. 管理接口组件610提供到配置组件605的接口,并与IP网络125通讯,从而与远程管理逻辑135通讯(见图1)。 Management interface component 610 provides an interface to configure component 605, the IP 125 and the communication network, thereby communicating with the remote management logic 135 (see FIG. 1). 在系统100中的每个处理器106包括处理器端存储逻辑的实例620。 Each processor in the system 100. Examples of processor 106 includes a memory logic 620 end. 每个那样的实例620经过2个RVI连接625与控制单点端存储逻辑615的对应实例通讯。 Examples of each of those 620 through two corresponding instances of the connection RVI communication control unit 625 and the end point 615 to store logic.

简而言之,配置组件605和接口610负责发现被分配到平台100的SAN存储器的那些部分,并负责使管理员将部分子分配到特定的PAN或处理器106。 Briefly, configuration component 605 and the interface 610 is responsible for discovering SAN are assigned to those portions of memory platform 100, and for enabling administrators to assign to a particular part of the sub-processor 106 or PAN. 存储配置逻辑605还负责将SAN存储分配通知控制节点端的逻辑615。 Configuration logic 605 is also responsible for storing the SAN storage allocation logic 615 notifies the control node side. 处理器存储逻辑620负责将通过内部互连110和存储的网115的处理器存储请求经专用RV I625通知控制节点端逻辑615。 The processor logic 620 is responsible for the storage interconnect grid 115 by 110 and stored in an internal node of the logic processor storage requests via a dedicated notification control 615 RV I625. 在某些实施例中该请求包含虚拟存储地址和SCSI命令。 In certain embodiments, the request includes a virtual storage address and SCSI command. 控制节点端逻辑负责通过识别对SAN的对应实际地址并通过将命令协议转换成对SAN适合的格式来接收和处理那样的命令,连接可以如光纤通道,但不限于此(带iSCSI的千兆位以太网是另外的示例性连接)。 The control node side is responsible for receiving and processing logic such as commands to identify the corresponding physical address and the SAN by SAN command protocol translate into a suitable format, such as Fiber Channel may be connected, but not limited thereto (with iSCSI Gigabit It is another exemplary Ethernet connection).

配置组件配置组件605确定,在SAN 130中哪个单元对每个单独的处理器106是可见的。 Determining configuration component configuration component 605, 130 in which cell SAN each individual processor 106 is visible. 它提供映射功能,通过他们附带的SCSI和光纤通道I/O接口128将处理器使用的设备号(如SCSI目标和LUN)翻译成对控制节点可见的设备号。 It provides a mapping function, through their accompanying SCSI and Fiber Channel I / O interface 128 to device number (e.g., SCSI target and LUN) processor for use visible translated to the control node device number. 它还提供访问控制功能,使处理器不能访问附着于该控制节点但不包括在该处理器和配置中的外部存储设备。 It also provides access control function, so that the processor can not access the control node is attached to but not including the external storage device and the processor configuration. 提供给处理器的(且给系统管理员及该处理器上的应用程序/用户)的模型使得它看起来似乎每个处理器具有附着于该处理器的接口的它自己的海量存储设备。 Provided to the processor (and to the application / system administrators and users on the processor) model makes it seem attached to each processor has its own mass storage device interface to the processor.

除了其他事情,此功能使得在一个处理器106上的软件容易地移到另外处理器。 Among other things, this feature allows software on a processor 106 easily moved to another processor. 例如在某些实施例,控制节点通过软件(不必任何物理上重新布电缆)可改变PAN的配置,使新的处理器能访问需要的设备。 In certain example embodiments, the control node of the PAN can be changed by software configuration (not necessarily physically rewiring any cable), so that the new device requires access to the processor. 因此能使新的处理器继承另一处理器的存储个性。 Thus enabling the new processor inherited personality to another storage processor.

在某些实施例中,控制节点看来象SAN上的主机,虽然另选的实施例允许处理器起那样的作用。 In certain embodiments, the control node appears as a host on a SAN, although the embodiment as alternative embodiments allow effects from the processor.

如上描述,配置逻辑发现分配给平台100的SAN存储(例如在平台自举时)且此存储池随后由管理员分配。 As described above, the configuration logic is assigned to find the platform SAN storage 100 (e.g., when the platform bootstrap), and the storage pool is then assigned by the administrator. 若发现在以后激活,完成发现操作的控制节点将新的视图与以前的视图比较。 If it is found after the activation, to complete the operation of the control node will find a new view compared to the previous view. 新的可用的存储被加入可由管理员分配的存储池。 Newly available storage is added to the storage pool may be assigned by the administrator. 消失的未指定的分区从可对PAN分配的可用的存储池中去除。 Disappearance of unspecified partition is removed from the available storage pool may be allocated to the PAN. 消失的分区被赋予触发器错误报文。 The disappearance of the partition is assigned trigger error messages.

管理接口组件配置组件605使管理软件能访问和更新下述信息,它描述在对控制节点120可见的设备和对单独的处理器106可见的虚拟设备之间的设备映射。 Configuration management interface component assembly 605 so that the management software can access and update the following information, which describes mapping between the device control apparatus 120 and the node is visible on a separate processor 106 is visible a virtual device. 它还允许访问控制信息。 It also allows access control information. 此指定能由处理节点结合模拟的SCSI盘的标识符来识别,如由模拟控制器、电缆、单元,或逻辑单元号(LUN)的名称识别。 This identifier can be specified by the SCSI disk binding analog processing node to identify, such as by an analog controller, cable unit, or a logical unit number (LUN) name identification.

在某些实施例,接口组件610与配置组件合作收集并监测信息和统计数据,如:·完成的I/O操作的总数·传输的字节总数·完成的读操作的总数·完成的写操作的总数·进行I/O的总时间量处理器端存储逻辑协议的处理器端逻辑620作为主机适配器模块来实现,后者通过在处理器106的操作系统中提供低层虚拟接口来模拟SCSI子系统。 Embodiment, the interface component 610 and the components arranged to cooperate in certain embodiments to collect and monitor information and statistics, such as: · completed I / O operation of transporting the total number of the total number of total number of bytes read-completion-write operation is completed the processor of the logic processor total amount of time the end of the total number of stored logical protocol-for I / O adapter module 620 implemented as a host, which provides low-level virtual interface by the operating system to simulate the processor 106 of the SCSI subsystem . 处理器106是也能够用此虚拟接口发送SCSI I/O命令到控制节点120进行处理。 The processor 106 is also able to use this interface to send the virtual SCSI I / O commands to the control node 120 for processing.

在使用冗余控制节点120的实施例中,每个处理节点105包括每个控制节点120的逻辑620的一个实例。 In embodiments using redundant control node 120, each processing node 105 includes one example of the control logic 620 of each node 120. 在某些实施例中,处理器使用物理设备编号而非逻辑号引用存储器。 In certain embodiments, the processor using physical device ID number rather than the logical memory reference. 即作为设备名规定地址,以识别LUN、SCSI目标、通道、主机适配器,和控制节点(如节点120a或120b)。 I.e. a predetermined address as a device name to identify LUN, SCSI target, channel, host adapter, and the control node (e.g., node 120a or 120b). 如图8所示,一个实施例将目标(T)和LUN(L)映射到主机适配器(H)、通道(C)、映射的目标(mT),和映射的LUN(ml)。 As shown, one embodiment of mapping the 8 target (T) and LUN (L) to the host adapter (H), the channel (C), the target (mT) mapping, and mapping the LUN (ml).

图7示出处理器端逻辑的示例性结构,逻辑720包括设备类型特定的驱动器(如磁盘驱动器)705。 Exemplary configuration FIG. 7 shows a processor of the logic, the logic 720 includes device-type specific driver (e.g., disk drive) 705. 中间层SCSI I/O驱动器710,和包装及互连逻辑715。 Intermediate layer SCSI I / O drivers 710, 715, and packaging and interconnect logic.

设备类型特定的驱动器705是操作系统提供的传统驱动器,并与特定的设备类型相关。 Device-type specific driver 705 is a conventional operating system drivers, and associated with a particular device type.

中间层SCSI I/O驱动器710是传统的中间层驱动器,一旦驱动器705判定设备是SCSI设备,中间层驱动器由设备类型特定的驱动器705调用。 Intermediate layer SCSI I / O driver 710 is a conventional intermediate layer driver, once the device driver 705 is a SCSI device determines the intermediate layer driver 705 is called by the device-type specific drive.

包装和互连逻辑715由中间层SCSI I/O驱动器710调用。 Packaging and interconnect logic 715 O driver 710 is invoked by the intermediate layer SCSI I /. 此逻辑提供SCSI子系统接口并因此模拟SCSI子系统。 This logic provides the SCSI subsystem interface and thus simulate the SCSI subsystem. 在某些是也使用千兆网结构的实施例中,逻辑715负责按需要包装SCSI命令,并负责与千兆网和RCLAN接口,使得NIC如上所述地通过到控制节点的专用的RVI发送包到控制节点。 In certain embodiments are also embodiments using Gigabit Ethernet configuration, the logic 715 is responsible for packaging as required SCSI command, and is responsible for the Gigabit Ethernet interfaces and RCLAN, so that the NIC dedicated RVI packet is transmitted to the control node as described above to the control node. 在下面上下文所述,修改对千兆网包的标题信息,以指出这是存储包并包括其它信息。 In the context of the following, modifications to the Gigabit Ethernet header information packet, to indicate that the packet is stored and include other information. 在图7中虽然未指出,包装逻辑715能使用RCLAN层支持并利用冗余的互连110及结构115。 In FIG 7, although not specified, the logic package 715 can be used to support and redundancy RCLAN layer interconnect structure 115 and 110.

对于使用千兆网结构115的实施例,连接725的RVI被指定1024个可用的VI的范围中的虚拟接口(VI)号。 For Example gigabit structure 115, the connection 725 is designated RVI number in the range of 1024 available VI Virtual Interface (VI). 对通讯的两个端点,交换机115用在这对端点(控制节点交换端口、控制节点VI号),(处理节点105交换端口、处理器节点VI号)之间的双向路径编程。 The two end points of communication, the switch 115 in this endpoint (port switching control node, the control node No. VI), a bidirectional path between the programmed (processing nodes exchange port 105, the processor node number VI).

对每类在任一方向发送的报文使用独立的RVI。 Packets of each type in any direction using a separate RVI. 因此总是有挂在每个RVI上的接收缓冲器,用于从协议的另一端发送的报文。 Therefore always hang on each RVI receive buffer for packets sent from the other end of the protocol. 此外,因为在每个RVI上只有一类报文在任一方向发送,连接到每个RVI通道的接收缓冲器按协议能使用的报文的最大报文长度合适地定尺寸。 Further, since only one class on each RVI packet transmitted in either direction, the maximum packet length of the packet channel connected to each receiving buffer RVI protocol can be used by suitably dimensioned. 在另外实施例中,所有可能的报文在单个RVI上多路复用,而不是使用两个VI。 In a further embodiment, all possible packets multiplexed on a single RVI, instead of two VI. 协议和报文格式不特定地需要使用两个RVI,且报文本身在其标题中具有报文类型信息,所以它们能被分离。 Protocol packet format and not specifically require the use of two the RVI, and the packets having a packet type information itself in its title, so that they can be separated.

两个通道之一用于交换SCSI命令(CMD)和状态(STAT)报文。 One of the two channels for the exchange of SCSI command (CMD) and a state (STAT) packets. 另一通道用于交换缓冲器(BUF)和发送(TRAN)报文。 Another channel used to exchange the buffer (BUF) and a transmission (TRAN) packets. 此通道也用于处理SCSI命令的数据有效载荷。 This channel is used for data processing the SCSI command payload.

CMD报文包含在节点105中的控制信息、要完成的SCSI命令、虚拟地址和I/O缓冲器的大小。 CMD packet control information contained in the node 105, to complete the SCSI command, and the virtual address I / O buffer size. STAT报文包含控制信息和反映在处理SCSI命令时已经发生的错误的完成状态码。 STAT message contains the error control code information and completion status is reflected in the SCSI command processing has occurred. BUF报文包含在控制节点120中控制信息和虚拟地址以及I/O缓冲器的大小。 BUF packet contains control information and the virtual address and the size of I / O buffer 120 in the control node. TRAN报文包含控制信息,并用于确认从节点105到控制节120的数据成功发送。 TRAN packet contains control information, and for confirming a successful data transmission to the node 105 from the control section 120.

处理器端的包装逻辑715检查要发送的SCSI命令以判定命令是否需要传输数据,若是,哪个方向传输。 Packaging Check the SCSI logical processor 715 to send a command end command to determine whether to transmit data, and if so, in which direction the transmission. 据此分析,包装逻辑715在报文标题中设置适当的标志信息。 According to this analysis, packaging logic 715 to set the appropriate flag information in the message header. 描述控制节点端逻辑的部分描述如何使用标志信息。 Description of the control node of the logic section describes how to use flag information.

在本发明的某些实施例中,在处理器端存储逻辑720和控制节点存储逻辑715之间的链路725能被用于传递控制报文,但不是SCSI协议的部分且不与SAN 130通讯。 In some embodiments part of the present invention, the link between the processor 715 and the control terminal of the storage node logic 720 store logic 725 can be used to transfer control messages, but not the SCSI protocol does not communicate with the SAN 130 . 相反,这些控制报文由控制节点端逻辑715处理。 Instead, these control packets are processed by the control logic 715 terminal node.

协议控制报文总是由协议的处理器端产生,并经过将处理器端逻辑720连接到控制节点端存储逻辑715的两个虚拟接口(VI)之一发送到协议的控制节点端。 Protocol control packet is always generated by the protocol processor side and connected to the processor 720 via the control node of the logic to the two ends of the logical storage 715 virtual interface (VI) transmits to the control node, one terminal protocol. 用于协议控制操作的报文标题与命令报文标题相同,差别在于使用不同的标志位区分作为协议控制报文的报文。 Protocol control message header for the operation command message header is similar to that using the flag bits of different control packets as packets. 控制节点120完成请求的操作,并用由状态报文的使用同样的报文标题在RVI上作出响应。 The control node 120 to complete the requested operation, and the state with the use of the same message packet responds the title RVI. 以此方式,对不常使用的协议控制操作不需要独立的RVI。 In this manner, less frequently used protocol operation no separate control RVI.

在使用冗余控制节点的某些实施例中,处理器端逻辑720检测从发出的命令来的某些错误,作为响应重发该命令到另外控制节点。 In some embodiments using redundant control node, processor logic 720 detects an end to a command issued from some errors, the command in response to the additional retransmission control node. 能在中间层驱动器710中实现此重试。 This can be achieved retry in the drive 710 the intermediate layer.

控制节点端存储逻辑在某些实施例中,控制节点端存储逻辑715作为设备驱动器模块实现。 A logic control terminal of the storage node In certain embodiments, the control node 715 implemented as a logic terminal of the storage device driver module. 逻辑715提供在控制节点120上对操作系统的设备层接口。 Logical layer interface providing apparatus 715 to the operating system on the control node 120. 此设备层接口用于访问配置组件705。 This layer interface device used to access configuration component 705. 在此设备驱动器模块初始化时,它响应从平台100中的所有处理器106来的协议报文。 When the device driver initialization module, responsive to the processor 106, all packets from the platform 100. 所有配置活动通过设备层接口引入。 All active interfaces introduced through the device layer. 所有I/O活动通过经互连110和交换结构115发送和接收的报文引入。 All I / O activity is introduced by interconnection through packet switch fabric 110 and 115 are sent and received. 在控制节点120上,每个处理器节点将有逻辑715的一个实例(虽然在图7中只显示一个框)。 The control node 120, each processor node 715 will have a logical instance (although only one display block in FIG. 7). 在某些实施例中,控制端逻辑715经过FCP或FCP-2协议或iSCSI或其他在各种媒体上使用SCSI-2或SCSI-3命令的其他协议与SAN 130通讯。 In certain embodiments, the control logic 715 via terminal FCP or FCP-2 protocol, or other protocols using iSCSI or other SCSI-2 or SCSI-3 command in various media and communication SAN 130.

如上所述,处理器端逻辑在RVI报文标题中设备标志,指出数据流是否与命令有关,若是,以什么方向。 As noted above, the processor side logic device RVI sign in the message header, indicates whether the data stream associated with the command, and if so, in what direction. 控制节点端存储逻辑715从处理器端逻辑接收报文,并随后分析标题信息以确定如何动作,如分配缓冲器等。 Terminal storing a logical control node 715 receives packets from the processor side logic, and header information is then analyzed to determine how operation, dispensing buffer or the like. 此外,该逻辑将包含在报文中的地址信息从处理器地址翻译到对应的映射的SAN地址,并发命令(如经过FCP或FCP-2)到SAN 130。 In addition, the logic will address information contained in the packet from the processor address translation map corresponding to a SAN address, and transmits a command (e.g., via FCP or FCP-2) to the SAN 130.

如TEST UNIT READY那样不需要SCSI数据传输阶段的SCSI命令由在RVI上发送用作命令报文的单条命令的处理器端逻辑720处理,并由在同一RVI上发回单个状态报文的控制节点端逻辑处理。 The TEST UNIT READY does not require that the data transmission phase SCSI command by the SCSI command on transmitted as a single packet RVI command processor 720 of the logic processing, by a single state back into the control node in the same message RVI logic processing ends. 更具体说,协议的处理器端用标准报文标题、对此命令的新序列号,希望的SCSI目标和LUN,要执行的SCSI命令,和零的列表长度来构筑报文。 More specifically, the agreement end processor with a standard message header, the new serial number for this command, hope SCSI target and LUN, SCSI commands to be executed, and to build a list of zero length packet. 逻辑的控制节点端接收报文,抽取SCSI命令信息并通过接口128将其传递给SAN 130。 The control node of the logic of the received message, extracts information and SCSI commands to pass 128 to SAN 130 via its interface. 在控制节点收到命令完成的回叫后,它使用标准的报文标题、对此命令的序列号、所完成的命令的状态,以及若命令带着校验条件状态完成还可选地使用请求辨别数据来构筑给处理器的状态报文。 After the node receives a control command completion callback, it uses standard message header, the sequence number of this command, the command completed status, and if the command is completed with the check condition status request also optionally be used identify data packets to build a state of the processor.

如READ命令那样需要从SCSI设备传输数据到主存储器的SCSI数据传输阶段的SCSI命令由处理器端逻辑处理,它发送命令报文到控制节点端逻辑715发送带着一个或RDMA WRITE操作的控制节点的回答到处理器节点105的存储器,并从控制节点端逻辑发送单个状态报文。 As required SCSI data transfer device to the main memory from the data transmission phase SCSI command by the SCSI READ command processor of the logic processing, it sends a command message to the control node 715 transmits a control node of the logic or with a RDMA WRITE operation answer memory 105 of processor node, and sending from the control node of the logic state of a single packet. 更具体说来,处理器端逻辑720用标准的报文标题、对此命令的新序列号、需要的SCSI目标及LUN、要执行的SCSI命令,和存储从命令来的数据的存储器区域的表来构筑命令报文。 More specifically, the table 720 of the logic processor with a standard message header, the new serial number to this command, the memory area required and LUNs SCSI target, SCSI command to be executed, and stores the command data to build a command packet. 当SCSI命令在控制节点执行时,控制节点端逻辑715分配暂时的存储器缓冲器来存储从SCSI操作来的数据。 When a SCSI command is executing the control node, the control node of the logic assigned temporary memory buffer 715 to store data from the SCSI operation. 在控制节点端逻辑715发送SCSI命令到SAN 130进行处理且该命令已完成之后,它用一个或多个RDMA WRITE操作的序列发回数据到处理器105的存储器。 After the end of the control node logic 715 SAN 130 to send a SCSI command and the command processing has been completed, it sends back the sequence data of one or more RDMA WRITE operations to the memory of the processor 105. 然后它用标准报文标题、对此命令的序列号,完成命令的状态构筑状态报文,若命令带着SCSI CHECKCONDITION状态完成时还可选地使用REQUEST SENSE数据。 It then uses a standard message header, the sequence number of this command, the command completion status messages constructed state, if the command is completed with the state SCSI CHECKCONDITION Shihai REQUEST SENSE data alternatively be used.

如WRITE命令那样需要SCSI数据传输阶段来将数据从主存储器传输到SCSI设备的SCSI命令由处理器端逻辑720处理,它发送单个命令到控制节点端逻辑715,从控制节点端逻辑715发送一个或多个BUF报文到处理器端逻辑,从处理器端逻辑发送一个或多个RDMA WRITE操作到控制节点的存储器,从处理器端逻辑发送一个或多个TRAN报文到控制节点端逻辑,并从控制节点端逻辑发回单个状态报文到处理器端逻辑。 As required SCSI WRITE command to the data transfer phase transfer data from the main memory to the SCSI devices SCSI command processing by the processor 720 of the logic, it sends a single command to the control node of the logic 715, a transmission logic 715 from the control node side or BUF packet to the plurality of the logic processor, transmitting the one or more RDMA WRITE operations to the memory control node, transmitting from the one or more processors of the logic TRAN packet to the control node of the logic of the logic from the processor, and from the control node of the logic state of a single packet is sent back to the processor-side logic. 使用BUF报文将控制节点中临时缓冲器存储器的位置通知处理器端存储逻辑并使用TRAN报文表示RDMA WRITE数据传输完成,这样做是由于在底层千兆网的网中缺乏RDMAREAD能力。 Using the node position of the temporary buffer memory BUF packet processor notifies the control terminal of the logic and memory use TRAN represents RDMA WRITE packet data transfer is completed, this is due to lack of capacity in the network RDMAREAD bottom of Gigabit Ethernet. 若底层结构支持RDMA READ操作,能利用相应动作的不同顺列。 If the underlying structure supports RDMA READ operation, the corresponding operation can use different inline. 更具体地说,处理器端逻辑720用标准的报文标题、对此命令的序列号、希望的SCSI目标和LUN,和要执行的SCSI命令构造CMD报文。 More specifically, the processor 720 of the logic of a standard message header, the sequence number for this command, and the desired SCSI target LUNs, and SCSI command to execute CMD message configuration. 当SCSI命令在控制节点执行时,控制节点端逻辑715分配临时的存储器缓冲器来存储从SCSI操作来的数据。 When a SCSI command is executing the control node, the control node of the logic 715 allocates a temporary memory buffer to store the data from the SCSI operation. 然后协议的控制节点端用标准的报文标题、对此命令的序列号,和用于在该控制节点上临时存储器缓冲器的虚拟存储器的区域表来构筑BUF报文。 Control then end node protocol using standard message header, this command sequence number, and a virtual storage area table in the control node to construct a temporary memory buffer BUF packet. 然后处理器端逻辑720用一系列一个或多个RDMA WRITE操作发送数据到控制节点的存储器。 The processor logic 720 then ends a series of one or more RDMA WRITE operation of transmitting data to the control memory node. 它然后用标准报文标题,和对此命令的序列号构筑TRAN报文。 It then build TRAN message with a standard message header, and the serial number of this command. 在控制节点端逻辑发送SCSI命令到SAN 130进行处理,且接收到命令完成信号之后,它用标准的报文标题、对此命令的序列号、所完成的命令的状态构筑STAT报文,若命令完成带着CHECK CONDITION状态,还可选地使用REQUEST SENSE数据。 After the end of the control logic sends SCSI command to the node SAN 130 for processing, and a command complete signal is received, its sequence number state with a standard message header, this command, the command is completed build STAT message, if the command complete with CHECK cONDITION status, also optionally be used REQUEST SENSE data.

在某些实施例中,CMD报文包含存储命令的数据的虚拟存储器的区域表。 In certain embodiments, CMD packet contains a virtual memory area table data storage commands. BUF和TRAN报文还包含索引字段,它允许协议的控制节点端对在CMD报文中区域表中的每个条目发送独立的BUF报文。 TRAN and BUF packet further includes an index field, which allows the control node sends the protocol independent BUF packets for each entry packet CMD region table. 协议的处理器端通过对在BUF报文中描述的大量数据完成RDMA WRITE操作来回答那样的报文。 Side protocol processor by completing RDMA WRITE operations on large amounts of data described in the BUF packet as a reply message to. 随后通过TRAN报文指出单段数据传输的完成。 Then it pointed out that to complete the single piece of data packets transmitted by TRAN.

在处理器端逻辑720和控制节点端逻辑715之间的协议考虑分散-集中I/O操作。 In the protocol between the processor 715 and the control node 720 of the logic of the logic of dispersibility - centralized I / O operations. 此功能允许涉及I/O请求的数据能写入虚拟的和/或物理的存储器的若干不同的区域或从中读取。 This feature allows the data relates to I / O request can be written to the virtual and / or several different regions of physical memory or read therefrom. 这使在控制节点上多个非连续的缓冲器能用于该请求。 This allows multiple non-contiguous buffers can be used in the request to the control node.

如上所述,配置逻辑705负责发现分配给平台的SAN存储器,并与接口逻辑710连接,使得管理员能子分配存储器给特定的PAN。 As described above, configuration logic 705 is responsible for discovering the memory allocated to the SAN platform and connected to the interface logic 710, so that an administrator can assign the memory to a particular sub-PAN. 作为此分配的一部分,配置组件705建立并维护包括识别处理器地址和实际SAN地址之间对应关系的信息的存储数据结构915。 As part of this assignment, configuration component 705 includes information to establish and maintain a relationship between the recognition processor corresponding to the real address and store data structures 915 SAN address. 图7示出那样的结构。 Figure 7 shows such a structure. 如上所述,对应关系可以是在处理节点和模拟的SCSI盘的标识符之间的,标识符可以是如通过模拟控制器的名称、电缆、单元,或逻辑单元号的号。 As described above, the correspondence relation between the identifier may be a SCSI disk and analog processing node, the identifier may be a name, such as by resolution analog controller, cable unit, or a logical unit number.

管理逻辑管理逻辑135用于与控制节点软件连接以提供PAN。 Management logic management logic 135 is connected to the control node for the software to provide the PAN. 除其它事情,逻辑135使管理员建立PAN的虚拟网络拓扑,它对外部网的可见性(如作为服务群集),并建立在PAN上的设备类型,如网桥和路由器。 Among other things, logic 135 enables an administrator to create a virtual network topology PAN, and the visibility of its external network (such as a service cluster), and to establish the type of device on the PAN, such as bridges and routers.

逻辑135还与存储管理接口逻辑710连接,使得管理员在初始分配或其后能定义PAN的存储器。 Logic 135 is also connected with a storage manager 710 interface logic, so that an administrator can define the initial memory allocation PAN or after. 配置定义包括上面讨论的存储对应关系(SCSI对SAN的关系)和访问控制许可。 Configuration comprising a defined correspondence relationship storage (SCSI relation to the SAN) discussed above, and access control permissions.

如上讨论,每个PAN和每个处理器通过其虚拟连网(包括虚拟MAC地址)和虚拟存储器具有确定的个性。 As discussed above, each processor by each PAN and its virtual networking (including virtual MAC address), and a virtual memory having a determined character. 如下面讨论,记录那样个性的结构能由管理逻辑访问,以实现处理器的群集化。 As discussed below, such as the structure of the recording can be accessed by the individual management logic for the cluster of processors. 此外如上所述它们能被管理员访问,或由代理程序的管理员访问。 Also as mentioned above they can be administrator access, or access by the administrator of the agent. 例如,能响应如每天或每年的时间的某些事件使用代理程序重配置PAN,或响应系统上的某些负载而重配置。 For example, in response to certain events such as the time of day or year of using the PAN reconfiguration agent, or in response to some of the load on the system reconfiguration.

处理器上操作系统软件包括在千兆网交换器115上将控制台I/O通讯量路由到在控制节点上运行的管理软件的串行控制台驱动器码。 Operating system software on the processor 115 includes on the console I / O traffic routing switch Gigabit Ethernet driver code to the serial console running on a control node of the management software. 从那里管理软件使任何节点的控制台I/O流能通过控制节点的管理端口(其低速以太网端口机器紧急管理端口)或通过高速外部网125访问。 From there management software so that the management console port I / O stream by any node of the control node (low-speed Ethernet port emergency management port machine) or high-speed access to the external network 125. 为审查及记载历史的目的控制台的通信量能被记录。 As traffic review and chronicling the history of the purpose of the console can be recorded.

群集管理逻辑图9示出某些实施例的群集管理逻辑。 Cluster management logic 9 illustrates some embodiments of the cluster management logic. 群集管理逻辑905访问记录上述网络信息的数据结构910,如PAN的网络拓扑结构,在PAN中的MAC地址指定等。 Cluster management logic 905 to access the network information record data structure 910, such as the PAN network topology, MAC address specified in the PAN and the like. 此外,群集管理逻辑905访问记录各个处理器106的存储对应关系的数据结构915。 Further, the cluster management logic 905 to access record data structure 915 stores the corresponding relationship between each of the processor 106. 此外,群集管理逻辑905访问记录如平台100中未分配的服务器那样的空闲资源的数据结构920。 Further, the cluster management logic 905 access records in a data structure 100 such as the internet unallocated free resources server 920.

响应处理器的错误事件或管理员命令,群集管理逻辑905能改变数据结构,使给定处理器的存储和组网个性能“迁移”到新的处理器。 Event response processor error or administrator commands the cluster management logic 905 can change the data structure of a given processor storage and networking personality can "migrate" to the new processor. 新的处理器以这样方式“继承”以前处理器的个性。 The new processor "inheritance" in such manner as the previous processor personality. 群集管理逻辑905能这样做,将新的处理器交换到PAN以替代故障的处理器。 Cluster management logic 905 to do so, the new processor to exchange PAN to replace the failed processor.

新的处理器继承以前处理器的MAC地址并象以前那样工作。 The new processor inherited the previous processor MAC address and work as before. 当新处理器自举时控制节点通知连接性信息,并在必要时为无故障的处理器更新连接信息。 When the new control node bootstrap processor notifies the connection information, and updates the connection information when necessary, the processor is fault-free. 例如在某些实施例中,对其它处理器的RVI连接被透明地更新,即在其它处理器的软件不需要牵涉到建立到新交换入的处理器的连接。 For example, in certain embodiments, the processor is connected to the other RVI is updated transparently, i.e. it does not need to establish a connection involving the exchange to the new software in the processor of the other processor. 此外,新处理器继承以前的存储对应关系,因而继承以前处理器所维持的状态。 In addition, the new processor inherited previously stored correspondence relationship, and thus inherit the previous processor state maintained.

其它优点还有,允许有包括处理器的空闲资源的池,为整个平台共享,而不是只为给定的PAN共享。 There are other advantages, allowing free resources including processor pool, shared for the entire platform, not only for a given PAN sharing. 以此方式,能更有效地使用空闲资源(能保留来改善系统的可靠性及故障宽容度)。 In this way, more efficient use of idle resources (energy retention to improve system reliability and fault tolerance).

当新的处理器被“交换入”,它需要重新ARP,以便掌握IP地址对MAC地址的联系。 When the new processor is "switched in", it needs to re-ARP, in order to grasp the IP address of the MAC address of the contact.

另选方案当交换结构115的每个千兆网端口能支持其上的1024个同时虚拟接口连接,并用硬件保护保持它们互相隔离,操作系统能安全地与应用程序共享节点的千兆网端口。 Each alternative to Gigabit Ethernet ports when the switch fabric 115 can support the virtual interface 1024 while it is on, and by keeping them isolated from each other hardware protection, operating system and applications can be safely shared node Gigabit Ethernet port. 这就允许在应用程序之间的直接连接,而不需要通过整个驱动器码运行。 This allows direct connection between the application, without having to run through the entire driver code. 为此,操作系统的调用将建立虚拟接口通道并将其缓冲器和队列存储器映射到应用程序的地址空间。 To do this, call the operating system will create a virtual interface channels and memory buffers and queues are mapped to the application's address space. 此外,封装接口到通道的低层细节的库方便了那样虚拟接口连接的使用。 In addition, the package interface channel to facilitate low-level details of the library using as a virtual interface. 该库也能自动地建立冗余虚拟接口通道对,并在它们之间管理共享及故障在线恢复,而不需要调用应用程序的任何努力或了解。 The library can automatically build a redundant virtual interface channel pair, share and fault management and recovery online, without the need to call any effort to understand or applications between them.

上述实施例模拟内部通过类似ATM结构的以太网。 Example inside the ATM-like structure is simulated by Ethernet. 该设计能改变成使用内部以太网结构,这很大地简化了结构,如减轻了对模拟特征的需要。 This design change to use the internal Ethernet structure, which greatly simplifies the structure, such as reducing the need for analog behavior. 若外部网按ATM通讯,另外的变种是使用内部不必模拟以太网的ATM,且ATM在外部与那样编址的外部网通讯。 If the external network by ATM communications, another variant is not necessary to use an internal simulation of ATM Ethernet, and ATM on the outside with an external network communication, as addressed. 另外变种是在平台内部允许ATM(即不模拟以太网),只有外部通讯转换到以太网。 In addition variant is within the platform allows ATM (ie, not analog Ethernet), only external communications to Ethernet conversion. 这将是流线式内部通讯,但在控制器处需要模拟逻辑。 It will be streamlined internal communications, but the required analog logic controller.

某些实施例根据软件配置命令配置PAN。 PAN configured in accordance with certain embodiments the software configuration command. 可以看到,配置是根据程序控制。 Can be seen, it is arranged in accordance with program control. 例如,对那个PAN在高峰运行期间在软件控制下能配置更多的处理器,或在软件算法控制下对PAN能配置对应的更多或更少的存储空间。 For example, during peak periods of the PAN can be arranged more processors operating under software control, or corresponding to the configuration of the PAN can be more or less storage space under the control of software algorithms.

可以看到,本发明的范围不限于上述实施例,而是由附后的权利要求确定;这些权利要求应包容已经描述内容的修改及改进。 It can be seen, the scope of the present invention is not limited to the above-described embodiments, but determined by the claims appended hereto; These claims should accommodate modifications and improvements have been described with the contents.

Claims (58)

1.在具有多个计算机处理器、一个交换结构和到诸处理器的点对点链路的平台中模拟交换的以太网局域网的方法,包括:提供至少在两个计算机处理器上执行的以太网驱动器模拟逻辑;提供至少在一个计算机处理器上执行的交换器模拟逻辑;在交换器模拟逻辑和具有执行的以太网驱动器模拟逻辑的每个计算机处理器之间建立虚拟接口以允许它们之间的软件通讯,其中每个虚拟接口定义从一个计算机处理器到另一个计算机处理器经过交换结构的软件通讯路径;在具有执行的以太网驱动器模拟逻辑的每个计算机处理器和具有执行的以太网驱动器模拟逻辑的另一个计算机处理器之间建立虚拟的接口;若在一个计算机处理器和另一个之间满足预定准则地运行,一个计算机处理器的以太网驱动器模拟逻辑通过定义它们之间的软件通讯路径与另外的计算机处理 A method in an Ethernet LAN platform having a plurality of computer processors, a switch fabric and to point link in an analog processor such exchange, comprising: providing an Ethernet driver executing on the at least two computer processor analog logic; providing an analog switch logic executing on at least one computer processor; establishing a virtual interface between the logic and analog switch each Ethernet driver having a computer processor executing logic simulation software to allow therebetween communications, wherein each virtual processor interface definition from one computer to another computer via a processor communication path switching software configuration; each Ethernet driver having a computer processor executing logic simulation and the Ethernet driver has performed simulations establishing a virtual logical interface between the other computer processor; if predetermined criterion is met between a computer processor and another operating, Ethernet driver logic simulation by a computer processor to define a software communication path therebetween additional computer processing 单播通讯;以及若在一个计算机处理器和另一个之间不满足预定准则地运行,一个计算机处理器的以太网驱动器模拟逻辑通过连到发送单播通讯到另外计算机处理的交换模拟逻辑的虚拟接口与另外计算机处理器单播通讯。 Unicast communication; and if the predetermined criterion is not satisfied between a computer processor and another operating, Ethernet driver logic simulation by a computer processor coupled to the transmit unicast communication to another computer process simulation logical virtual exchange Interface with another computer processor unicast communication.
2.如权利要求1的方法,其特征在于具有执行的以太网驱动器模拟逻辑的每个计算机处理器与一个虚拟的MAC地址相关,且其中MAC地址按规则形成,将计算机处理器识别成与外部网络的MAC地址不同的多个计算机处理器之一。 2. A method as claimed in claim 1, characterized in that each of the Ethernet driver having a computer processor executing logic simulation associated with a virtual MAC address, and wherein the MAC address is formed by the rules, the processor identifies the computer to the external one of a plurality of different computer processors MAC address of a network.
3.如权利要求2的方法,其特征在于平台通过用于与外部网络通讯的接口逻辑连接到外部网络,其中外部网络接口逻辑与它自己的MAC地址相关,且其中在外部网络上使用外部网络接口逻辑的MAC地址通讯报文。 3. A method as claimed in claim 2, characterized in that the internet connected through a communication network to the external interface logic to the external network, wherein the external network interface logic associated with its own MAC address, and wherein the external network to an external network MAC address of the interface logic of communication packets.
4.如权利要求1的方法,其特征在于所述多个计算机处理器中的第一计算机处理器使用第一虚拟接口与所述多个计算机处理器中的第二计算机处理器单播通讯,而其中第二计算机处理器使用不同虚拟接口通讯到第一计算机处理器。 4. A method as claimed in claim 1, wherein said first plurality of computer processor computer processor using the first virtual interface and the second plurality of computer processor computer processor unicast communication, and wherein the second virtual interface computer processors use different communication to the first computer processor.
5.如权利要求1的方法,其特征在于每个计算机处理器包括用于在点对点链路上通讯的交换结构驱动器逻辑且包括求校验和的能力,其中以太网驱动器模拟逻辑包括求校验和的能力,但当交换结构驱动器逻辑已对一报文求校验和时,禁止那样的求校验和。 5. A method as claimed in claim 1, characterized in that each of the computer processor includes a switch fabric driver logic on-point link communication capabilities and includes the checksum, wherein the Ethernet driver logic includes an analog Checksum and capacity, but logically drives the switch fabric to the packet when a checksum, the checksum that is prohibited.
6.如权利要求5的方法,其特征在于交换结构驱动器逻辑实现可靠的通讯协议,以保证在交换结构上接收报文。 6. The method as claimed in claim 5, wherein the switch fabric driver logic reliable communication protocol, to ensure that the received packet in the switch fabric.
7.如权利要求1的方法,其特征在于交换结构和点对点通讯链路以冗余配置方式安排。 7. A method as claimed in claim 1, characterized in that the switching point to point communication link configuration and a redundant configuration arranged.
8.如权利要求1的方法,其特征在于以太网驱动器模拟逻辑通过经虚拟接口发送报文到交换模拟逻辑来广播通讯一报文,且其中交换模拟逻辑从虚拟接口接收和克隆广播报文,且发送克隆的报文到网络中其它计算机处理器。 8. The method as claimed in claim 1, characterized in that the Ethernet driver analog logic to interface to send packets via virtual switching logic in the analog broadcast to a communication packet, and wherein the switching from the virtual logical analog interface for receiving broadcast packets and cloning, cloning and send packets to the network of other computers processors.
9.如权利要求1的方法,其特征在于交换模拟逻辑对模拟的网络定义并维护计算机处理器成员。 9. The method as claimed in claim 1, characterized in that the switching network definition analog and analog logic processor computer maintenance member.
10.如权利要求1的方法,其特征在于以太网驱动器模拟逻辑发送大于最大发送单位MTU长度的报文。 10. The method as claimed in claim 1, characterized in that the Ethernet driver analog logic transmits packets larger than the maximum transmission unit MTU length.
11.用于模拟交换以太局域网的系统,包括:多个计算机处理器;到这些处理器的交换结构和点对点链路;在交换结构及点对点链路上建立虚拟接口的虚拟接口逻辑,其中每个虚拟接口定义从计算机处理器经交换结构到另外计算机处理器的软件通讯链路径;在至少二个计算机处理器上执行的以太网驱动器模拟逻辑;在至少一个计算机处理器上执行的交换模拟逻辑,包括:建立在交换模拟逻辑和具有执行的以太网驱动器模拟逻辑之间的虚拟接口,使在其间能作软件通讯的逻辑;接收从一个虚拟接口到具有执行的以太网驱动器模拟逻辑的计算机处理器的报文,并发送报文到具有执行的以太网驱动器模拟逻辑的另外的计算机处理器,以响应与该报文相关的地址信息的逻辑;及建立在每个具有执行的以太网驱动器模拟逻辑的计算机处理器和每个另外的具有执 11. A system for simulating switched Ethernet LAN, comprising: a plurality of computer processors; point link to the switch fabric, and these processors; virtual interface to create a virtual interface logic in the switch fabric and point to point link, wherein each virtual interface software communications link path is defined from the computer processor is switched to another configuration of the computer processor; Ethernet driver logic simulation is performed on at least two computer processor; switched analog logic executing on at least one computer processor, comprising: establishing a logical exchange having an analog interface between the virtual Ethernet driver performs the analog logic, and the logic can be in the software communication therebetween; from a computer processor having a virtual interface to Ethernet driver performs the reception analog logic message, and transmits the message to another computer having a processor, an Ethernet driver logic simulation is performed in response to the logical address information associated with the packet; and establishing Ethernet driver logic simulation performed each having the computer processor and each further having performed 的以太网驱动器模拟逻辑的计算机处理器之间的虚拟接口的逻辑;其中以太网驱动器模拟逻辑包括与模拟的以太网中另外计算机处理器单播通讯的逻辑,若虚拟接口满足预定准则地运行,通过定义它们之间软件通讯路径的虚拟接口;若虚拟接口不满足预定准则地运行,则通过交换模拟逻辑。 Logical Virtual Ethernet driver interface between the analog logic of the computer processor; wherein the Ethernet driver logic includes logic and analog simulation computer processor additionally Ethernet unicast communication, if the virtual interface satisfies a predetermined criterion to run, by defining a virtual interface software communication path therebetween; it does not meet the predetermined criterion if the virtual interface operates, through switching logic simulation.
12.如权利要求11的系统,其特征在于具有执行的以太网驱动器模拟逻辑的每个计算机处理器与虚拟MAC地址相关,且其中MAC地址按规则形成,将计算机处理器识别成多个由外部网络的MAC地址区分的计算机处理器之一。 12. The system of claim 11, wherein each computer processor having Ethernet driver performs the analog logic associated with the virtual MAC address, and wherein the MAC address is formed by the rules, a computer processor identifies a plurality of external MAC address of the computer to distinguish between one network processor.
13.如权利要求12的系统,其特征在于还包括用于与外部网络通讯的外部网络接口逻辑,其中外部网络接口逻辑与它的MAC地址相关,且其中交换模拟逻辑包括用于发送报文到用于在外部网络上通讯的外部网络接口逻辑的逻辑,其中那样的报文使用外部网络接口逻辑的MAC地址在外部网络上通讯。 13. The system claimed in claim 12, characterized by further comprising an external network interface logic for communication with the external network, wherein the external network interface logic associated with its MAC address, and wherein the switching logic comprises means for transmitting the analog packet to the network interface logic for the external logic on the external network communication, wherein packets as external network interface logic in the MAC address on the external communications network.
14.如权利要求11的系统,其特征在于所述多个计算机处理器中的第一计算机处理器使用第一虚拟接口与所述多个计算机处理器中的第二计算机处理器单播通讯,但其中第二计算机处理器使用不同的虚拟接口通讯到第一计算机处理器。 14. The system of claim 11, wherein said first plurality of computer processor computer processor using the first virtual interface and the second plurality of computer processor computer processor unicast communication, but using a second computer processor can correspond to a different virtual processor of the first computer.
15.如权利要求11的系统,其特征在于每个计算机处理器包括用于在点对点链路上通讯的交换结构驱动器逻辑,且其中交换结构驱动器逻辑包括求校验和的能力,且其中以太网驱动器模拟逻辑包括求校验和的能力且包括在交换结构驱动器逻辑已对报文作了校验和时在以太网驱动器模拟逻辑中禁止求校验和的逻辑。 15. The system of claim 11, wherein each computer processor includes a switch fabric driver logic link on a point to point communication, and wherein the switch fabric includes a driver logic capability checksum, and wherein the Ethernet analog driver logic includes the ability to include a checksum in the switch fabric and driver logic have been made in the checksum logic prohibition Ethernet driver logic simulation checksum of the packet.
16.如权利要求15的系统,其特征在于交换结构驱动器逻辑实现可靠的通讯协议,以保证在交换结构上接收报文。 16. The system of claim 15, wherein the switch fabric driver logic reliable communication protocol, to ensure that the received packet in the switch fabric.
17.如权利要求11的系统,其特征在于交换结构和点对点链路以冗余配置方式安排。 17. The system as claimed in claim 11, characterized in that the switching point link structure and arranged in a redundant configuration.
18.如权利要求11的系统,其特征在于以太网驱动器模拟逻辑包括通过经虚拟接口发送报文到交换模拟逻辑而广播通讯一个报文的逻辑,且其中交换模拟逻辑包括从虚拟接口接收和克隆广播报文并发送克隆的报文到网络中其它计算机处理器的广播逻辑。 18. The system of claim 11, wherein the Ethernet driver logic includes an analog interface to send packets to the switching logic analog broadcast traffic through a packet via a virtual logic, analog logic, and wherein the exchanging comprises receiving from the virtual interface and cloning broadcast packets and sends the packets to the network clone broadcast logical processors of other computers.
19.如权利要求11的系统,其特征在于交换模拟逻辑包括定义并维护模拟网络的计算机处理器成员的逻辑。 19. The system as claimed in claim 11, characterized in that the switching logic simulation processor comprises logic members define and maintain a network of computer simulation.
20.如权利要求11的系统,其特征在于以太网驱动器模拟逻辑包括发送大于最大发送单元MUT尺寸的报文的逻辑。 20. The system as claimed in claim 11, characterized in that the Ethernet driver logic includes an analog logic transmission packets is greater than the maximum transmission unit size of the MUT.
21.在具有多个处理器的计算平台中实现地址解析协议ARP的方法,包括:定义在该计算平台上要模拟的以太网的拓扑结构,该拓扑结构包括处理器节点和交换节点;从多个处理器指定一组处理器作为处理器节点作用的处理器;指定一个处理器起交换节点的作用;对模拟的以太网的每个处理器节点指定虚拟MAC地址;在底层物理网络上分配虚拟接口,以提供从每个处理器节点到每个其它处理器节点的直接软件通讯,其中每个虚拟接口具有对应的标识符;向交换节点发送ARP请求的处理器节点,其中ARP请求包括IP地址;发送ARP请求到模拟的以太网中其他处理器节点的交换节点;与IP地址相关向交换节点发出ARP回答的处理器节点,回答包含与该IP地址相关的处理器节点的虚拟MAC地址;接收ARP回答并修改该ARP回答,使之包括对一虚拟接口的虚拟接口识别的交换节点 21. implemented in a computing platform having a plurality of processors in the address resolution protocol ARP method, comprising: defining on the computing platform to be simulated topology Ethernet topology the node comprising a processor and a switching node; from multiple processor specifies a set of processor nodes acting as a processor a processor; a processor from switching node designated role; specified for each virtual MAC address of the Ethernet analog processor node; virtual distribution on the underlying physical network interface to provide direct node from each processor to every other processor node communication software, wherein each virtual interface has a corresponding identifier; sending ARP requests to the processor node switching node, wherein the request includes an IP address of the ARP ; send ARP requests to the Ethernet switching node simulation of other processor nodes; associated with the IP address that processor node to switching node ARP reply, answer includes a processor node virtual MAC address associated with the IP address; receiving ARP reply and modify the ARP reply, to include the identified virtual interface of a virtual interface switching node 发出ARP请求的处理器节点将该虚拟接口用于与关于该IP地址的处理器节点的随后通讯。 The processor node sends this ARP request with a virtual interface for subsequent communication processor node about the IP addresses.
22.如权利要求21的方法,其特征在于底层物理网络是连接多个处理器的点对点网络。 22. The method as claimed in claim 21, characterized in that the underlying physical network is an ad hoc network connecting a plurality of processors.
23.如权利要求21的方法,其特征在于处理器的子集被组合成群集,且该群集中的一个处理器是负载均衡处理器节点,且其中当群集中的任何处理器发出ARP请求时,交换节点修改该ARP回答,使之包括对负载均衡处理器节点的虚拟接口标识符。 23. The method of claim 21, wherein the subset of processors are grouped into clusters, and the cluster load balancing processor is a processor node, and wherein when any processor sends an ARP request to the cluster when switching node to modify the ARP reply, to include the load balancing virtual interface identifier of the processor node.
24.如权利要求21的方法,其特征在于交换节点与外部IP网络通讯,且其中通讯ARP回答的动作包括识别该ARP回答是从平台的处理器节点而来。 24. The method of claim 21, wherein the switching node with the external IP network communication, and wherein the communication action comprise ARP reply identifying the ARP reply comes from the platform is a node processor.
25.地址解析协议ARP系统,包括:具有由底层物理网络连接的多个处理器的计算平台;在一个处理器上可执行以定义在计算平台上要模拟的以太网的拓扑结构的逻辑,该拓扑结构包括处理器节点和交换节点。 25. The ARP protocol system comprising: a computing platform connected to the network by a plurality of underlying physical processors; executable logic to define on the computing platform to be simulated topology Ethernet on a processor, topology including processor nodes and switching nodes. 在一个处理器上可执行以便从多个处理器指定一组处理器作为用作处理器节点的处理器的逻辑;在一个处理器上可执行以便对该模拟的以太网的每个处理器节点指定虚拟MAC地址的逻辑;在一个处理器上可执行以便在底层物理网络上分配虚拟接口使得提供从每个处理器节点到每个另外处理器节点的直接软件通讯的逻辑,其中每个虚拟接口具有对应的标识符;具有向交换节点通知ARP请求的ARP请求逻辑的每个处理器节点,其中ARP请求包括IP地址;包括将ARP请求通讯到模拟的以太网上的所有其它节点的ARP请求广播逻辑的交换节点;具有ARP回答逻辑的每个处理器节点。 Executable on a processor to specify a plurality of processors from a set of logical processors as a processor as the processor node; executable on a processor for each of the analog processor node Ethernet specifies a virtual logical MAC address; executable in a processor assigned to a virtual interface software that provides direct communication processor node from each node to each additional logic in the processor on the underlying physical network, wherein each virtual interface has a corresponding identifier; each processor node having logic ARP request ARP request notification to the switching node, wherein the ARP request including the IP address; ARP comprises all other nodes on the Ethernet ARP requests communication to the analog broadcast request logic switching nodes; each ARP reply processor node having logic. 回答逻辑判断它是否为与ARP请求中的IP地址相关的处理器节点,若是,向交换节点发出ARP回答,其中ARP回答还包含与该IP地址相关的处理器节点的MAC地址;包括ARP回答逻辑的交换节点,该逻辑接收ARP回答并修改ARP回答,使之包括对ARP请求的节点的虚拟接口标识符。 Answer logic determines whether it is associated with a processor node an IP address in the ARP request, and if so, an ARP reply sent to the switching node, further wherein an ARP reply contains the MAC address associated with that processor node IP address; ARP reply including logic the switching node, the logic receives ARP reply and modify ARP reply, to include virtual interface node identifier ARP request.
26.如权利要求25的系统,其特征在于底层物理网络是连接多个处理器的点对点网络。 26. The system as claimed in claim 25, characterized in that the underlying physical network is an ad hoc network connecting a plurality of processors.
27.如权利要求25的系统,其特征在于组织处理器的一个子集作为群集,且其中该群集中的一个处理器是负载均衡处理器节点,且其中交换节点包括检测从处理器节点来的ARP回答是否从群集中任何处理器来的逻辑,若是,则修改ARP回答使包括对负载均衡处理器节点的虚拟接口标识符。 27. The system of claim 25, wherein a subset of the tissue processor as a cluster, and wherein a processor of the cluster load balancing processor node, and wherein the switching node comprises a node processor detects from ARP answer whether any centralized logic processor from the group, and if so, to modify the ARP reply including virtual interface identifier load balancing processor nodes.
28.如权利要求25的系统,其特征在于交换节点与外部IP网络通讯,且其中处理器节点的ARP回答逻辑包括识别该ARP回答是从平台中的一个处理器节点来的逻辑。 28. The system of claim 25, wherein the switching node with the external IP network communication, and wherein the processor node ARP reply logic includes logic to identify the ARP reply from the platform of a processor node.
29.一个用于计算机处理的平台,包括:连接到内部通讯网络的多个计算机处理器;至少一个控制节点与外部通讯网络及具有外部存储地址空间的外部存储网络通讯,且其中至少一个控制节点连接到内部网络,从而与多个计算机处理器通讯;配置逻辑,它从多个处理器中定义并建立具有对应的计算机处理器组的虚拟处理区域网;定义并建立在该计算机处理器组中提供通讯但排除不在定义在组中的多个处理器的虚拟局域通讯网;定义并建立对存储网络的地址空间有定义的对应关系的虚拟存储空间。 29. A platform for a computer process, comprising: a plurality of computer processors coupled to the internal communications network; at least one control node and an external communication network and the external network communication with the external storage memory address space, and wherein the at least one control node connected to the internal network, thereby communicating with a plurality of computer processors; configuration logic, which defines the plurality of processors and processing to create a virtual local area network has a corresponding set computer processor; defining and establishing in the computer processor group provide communications excluding virtual local area communication network is not defined in the group of the plurality of processors; define and establish the address space of virtual storage network storage space defined correspondence relationship.
30.如权利要求29的平台,其特征在于控制节点通过内部通讯网络接收编址到外部通讯网络的实体的通讯报文,且其中控制节点包括在对应于已接收的报文的外部通讯网络上提供报文的逻辑。 30. The platform of claim 29, wherein the communication packets via the internal communication network control node receives communications addressed to an external network entity, and wherein the control node comprises on the external communication network corresponding to the received packet provide packet logic.
31.如权利要求29的平台,其特征在于控制节点通过外部通讯网络接收编址到平台上的实体的通讯报文,且其中控制节点包括将报文提供给对应于所接收的报文的编址实体的逻辑。 31. The platform of claim 29, wherein the control node through the packet communication entity on the external network reception communication addressed to the platform, and wherein the control node comprises a packet is supplied to the encoding packets corresponding to the received the logical address of the entity.
32.如权利要求29的平台,其特征在于计算机处理器和控制节点包括在内部通讯网络上模拟以太网功能的网络模拟逻辑。 32. The platform as claimed in claim 29, wherein the control node comprises a computer processor and network simulation logic simulation function on the Ethernet internal communications network.
33.如权利要求32的平台,其特征在于内部通讯网络是点对点交换结构。 33. The platform as claimed in claim 32, characterized in that the internal communication network is a peer switch fabric.
34.如权利要求29的平台,其特征在于内部通讯网络包括将计算机处理器和至少一个控制节点连接到冗余交换结构的冗余互连。 34. A platform as claimed in claim 29, characterized in that the internal communication network comprises a computer processor and connected to the at least one control node to a redundant switch fabric redundant interconnection.
35.如权利要求34的平台,其特征在于具有至少一个连接到该互连的其它控制节点,并形成冗余控制节点。 35. The platform as claimed in claim 34, characterized in that at least one other control nodes coupled to the interconnect, and a control node for redundancy.
36.如权利要求29的平台,其特征在于控制节点通过内部通讯网络接收从计算机处理器来的存储报文,且其中控制节点包括从接收的存储报文提取一地址,以识别在外部存储地址空间中定义的对应地址,并在对应于接收的存储报文并具有对应地址的外部存储网络上提供报文的逻辑。 36. The platform of claim 29, wherein the control node receives from a computer processor to store the internal communication network packets, and wherein the control node comprises extracting an address from the storing received packets to identify the external memory address corresponding to the address space defined, and corresponding to the received packets and store packets having logic provides the external memory address corresponding to a network.
37.如权利要求36的平台,其特征在于控制节点包括缓冲对应于从计算机节点接收的写报文的数据并将提供的对应报文中的缓冲数据提供给外部存储网络的逻辑。 37. The platform as claimed in claim 36, characterized in that the control node comprises a packet buffer corresponding to the write data received from the computer node and providing a buffer packet data corresponding to the supplied external logical storage network.
38.如权利要求36的平台,其特征在于控制节点从外部存储网络接收存储报文,且其中控制节点包括识别响应所接收的报文的对应计算机处理器或控制节点,并将对应的报文提供给识别的处理器或控制节点的逻辑。 38. A packet as claimed in claim 36 platform, wherein the control node stores the network received from the external packet memory, and wherein the control node comprises a control node corresponds to a computer or a processor in response to the received identification message, and the corresponding provided to a processor or control logic identifies the node.
39.部署虚拟处理区域网的方法,包括下列动作:提供具有连接到内部通讯网络的多个计算机处理器及至少一个控制节点的平台,其中控制节点与外部通讯网络及具有外部存储地址空间的外部存储网络通讯;对虚拟处理网络定义对应的计算机处理器组;建立提供在计算机处理器组中通讯的虚拟局域通讯网络,但在多个处理器中排除未定义在组中的处理器;定义在虚拟处理网络的虚拟存储空间与已定义的存储网络的地址空间之间的对应关系。 39. A method of deploying virtual processing area network, comprising the acts of: providing a platform connected to the internal communications network more computer processors and at least one control node, wherein the control node with the external communication network and the external memory having an external address space storing communication network; virtual processing corresponding to the definition of the network computer processor group; providing virtual local area communications networks to establish communications in a computer processor group, but excluding undefined group of processors in the plurality of processors; defined correspondence between the virtual memory space of the virtual address space processing network storage network has been defined.
40.如权利要求39的方法,其特征在于控制节点通过内部通讯网络接收编址到外部通讯网络的实体的通讯报文,且其中该控制节点在对应于所接收的报文的外部通讯网上提供报文。 40. The method of claim 39, wherein the control node receives a communication network via an internal communication packet addressed to the external communication network entity, and wherein the control node is provided corresponding to the received packets external communication line message.
41.如权利要求39的方法,其特征在于控制节点通过外部网络接收编址到在平台上的实体的通讯报文。 41. The method of claim 39, wherein the external network control node receives packets addressed to the communication entity on the platform. 并其中控制节点提供报文给对应于接收的报文的编址实体。 And wherein the control node to provide a packet corresponding to the received packets addressed entity.
42.如权利要求39的方法,其特征在于计算机处理器和控制节点在内部通讯网络上模拟以太网功能。 42. The method as claimed in claim 39, wherein the computer processor and control nodes on the internal analog functions Ethernet communication network.
43.如权利要求42的方法,其特征在于内部通讯网络是点对点的交换结构,且其中在内部点对点交换结构上提供以太网功能的模拟。 43. The method as claimed in claim 42, characterized in that the internal structure of communication network exchange point, and wherein a functional analog of Ethernet switching point on the internal structure.
44.如权利要求39的方法,其特征在于计算机处理器在连接计算机处理器及至少一个控制节点的冗余互连上通讯。 44. The method as claimed in claim 39, wherein the computer processor to communicate over a computer processor and a redundant interconnection connecting the at least one control node.
45.如权利要求44的方法,其特征在于具有至少一个连接到该互连的其它控制节点,并形成冗余的控制节点。 45. The method as claimed in claim 44, characterized in that a control node is connected to at least one of the other control nodes interconnected and form redundant.
46.如权利要求39的方法,其特征在于控制节点通过内部通讯网络接收从计算机处理器来的存储报文,且其中控制节点从接收的存储报文提取地址,识别在外部存储地址空间中的定义的对应地址,并在对应于接收的存储报文且具有对应地址的外部存储网络上提供报文。 46. ​​The method of claim 39, wherein the control node receives from the computer processor through the internal packet storing the communication network, and wherein the control node extracts the address from the storing received packets, identifying the external memory address space corresponding to the address defined and stored corresponding to the received message packet and having provided on the external memory corresponding to the network address.
47.如权利要求46的方法,其特征在于控制节点缓冲对应于从计算机处理器接收的写报文的数据,并提供缓冲的数据在向外部存储网络提交的对应报文中。 47. The method of claim 46, wherein the control node buffer corresponding to the received packet from the computer processor to write data, and provides data corresponding to the packet buffered in the external storage submitted to the network.
48.如权利要求46的方法,其特征在于控制节点从外部存储网络接收存储报文,且其中控制节点识别应答该接收的报文的计算机处理器或控制节点,并提供相应的报文给识别的处理器或控制节点。 48. The method of claim 46, wherein the control node stores the network received from the external packet memory, and wherein the control node identification transponder or a computer processor, the control node of the received packet and provides a corresponding message to the identification a processor or a control node.
49.可连接到外部通讯网络和存储网络的用于计算机处理的平台,包括:连接到内部通讯网络的多个计算机处理器;配置逻辑,用于定义并建立(a)在内部网络上的虚拟局域通讯网络,其中在虚拟局域通讯网络中的每个计算机处理器具有对应的虚拟MAC地址,且提供在一组计算机之间的通讯,但排除多个处理器中不在定义的组中的处理器,和(b)对该存储网络的地址空间有定义的对应关系的虚拟存储空间;故障在线恢复逻辑,响应计算机处理器的故障,从多个处理器分配一个计算机处理器以替代故障的处理器,故障在线恢复逻辑包括:将故障的处理器的MAC地址赋给替代故障处理器的处理器的逻辑;将故障处理器的虚拟存储空间和定义的对应关系赋给处理器的逻辑;和重建包括替代故障处理器的处理器并排除故障的处理器的虚拟局域网的逻辑。 49. The platform may be connected to a computer for processing external communication network and storage network, comprising: internal communication network connected to a plurality of computer processors; logic configured to define and establish (a) on the internal virtual network local communication network, wherein each computer processor virtual local area communication network having virtual MAC address and provides the communication between a group of computers, but the exclusion of the plurality of processors is not defined a processor, and (b) the storage address space of the virtual network storage space defined correspondence relationship; line fault recovery logic, in response to a failure of the computer processor, the computer processor to assign a plurality of processors to replace the failed processor, logic fault recovery line comprises: the MAC address of a processor assigned to replace the failed processor logical processor failures; the correspondence relationship of the virtual memory space and defined fault handler assigned to logical processors; and Alternatively reconstruction logic including a processor and a processor fault troubleshooting virtual local area network processor.
50.如权利要求49的平台,其特征在于配置逻辑建立虚拟接口以在虚拟网络的处理器中定义软件通讯路径,且故障在线恢复逻辑包括建立从在虚拟网络中的处理器到替代故障处理器的处理器的虚拟接口的逻辑。 50. The platform of claim 49, wherein the logic configured to establish a virtual interface is defined in a virtual network processor software communication path, and a fault recovery line comprises establishing from the logic processor in the virtual network to replace the failed processor the logical processors of a virtual interface.
51.如权利要求49的平台,其特征在于配置逻辑从第二组计算机处理器和与存储网络地址空间有定义对应关系的第二虚拟存储空间建立第二虚拟局域网,且其中故障在线恢复逻辑使替代故障的处理器的处理器继承故障处理器的虚拟局域网和虚拟存储的个性。 51. The platform of claim 49, wherein the logic configured from a second set of computer processors and a network address storage space definitions relate to a second virtual storage space to establish a second virtual local area network, and wherein the logic that the line fault recovery replace the faulty processor processor processor failure to inherit a virtual local area network and virtual storage personality.
52.在具有连接到内部通讯网络的多个计算机处理器的平台中进行计算机处理的方法:在内部网络上定义并建立虚拟局域通讯网,其中在虚拟局域通讯网中的每个计算机处理器具有对应的虚拟MAC地址,且虚拟局域网提供在一组计算机处理器之间的通讯,但排除多个处理器中不在定义的组中的处理器;用对存储网络的地址空间的已定义的对应关系定义并建立虚拟存储空间;响应一个计算机处理器的故障,从多个处理器中分配一个计算机处理器以替代故障的处理器,包括将故障的处理器的MAC地址赋给替代故障的处理器的处理器,将虚拟存储空间以及故障处理器的定义的对应关系赋给替代故障处理器的处理器,并重建虚拟局域网使包括替代故障处理器的处理器并排除故障的处理器。 52. The method of computer processing in a computer platform having a plurality of processors connected to the internal communication network in: defining and establishing a virtual local area network on the internal communications network, wherein each computer processor has a virtual local area communication network corresponding to the virtual MAC address, and a virtual local area network providing communications between a group of computer processors, but does not exclude a plurality of processors in the processor group defined above; with the correspondence relationship stored in the address space of the network defined define and establish a virtual storage space; response to a failure of a computer processor, a computer processor allocating the processor to the processor from the plurality of alternative fault, including the MAC address of the failed processor is assigned to replace the failed processor processor, a correspondence relationship defined by the virtual storage and a processor failing processor is assigned to replace the failed processor, and virtual local area network so that the reconstruction processor comprises a processor and replace the failed processor troubleshooting.
53.如权利要求52的方法,其特征在于在建立虚拟局域网时建立虚拟接口,以定义在虚拟网的诸处理器之间的软件通讯路径,并在一个处理器替代故障的处理器时,建立到替代故障的处理器的处理器的虚拟接口。 53. The method of claim 52, wherein establishing a virtual interface is creating virtual local area network, to define a communication path between the various software virtual network processor and the processor when a fault alternate processor to establish processor to replace the failed virtual interface processor.
54.如权利要求52的方法,其特征在于用第二组计算机处理器和与存储网络地址空间有定义的对应关系的第二虚拟存储空间建立第二虚拟局域网,且在处理器故障时,替代故障处理器的处理器继承故障处理器的虚拟局域网和虚拟存储的个性。 54. The method of claim 52, wherein a second set of computer processor and storage network address space and a second virtual storage space to establish a correspondence relationship defined by a second virtual local area network, and, when a processor failure, an alternative failing processor processor processor failure to inherit a virtual local area network and virtual storage personality.
55.提供由IP地址编址的服务的系统,包括:至少两个计算机处理器,每个包括提供服务的逻辑;用于接收对服务的请求报文的群集逻辑,该报文具有IP地址,并用于将请求分配到具有提供该服务的逻辑的至少两个计算机处理器中的一个。 55. The services provided by the system addressing IP address, comprising: at least two computer processor, including providing each service logic; receiving a request for a logical cluster to a service packet, the packet having an IP address, and for requesting to assign the at least two computer processor having a logic that provides the service in.
56.如权利要求55的系统,其特征在于用于分配的逻辑包括在确定哪个处理器服务该报文时分析在输入报文中的源信息的逻辑。 56. The system of claim 55, wherein the distribution logic includes logic for analyzing the incoming packets in the packet service when the processor determines which source information.
57.提供由IP地址编址的服务的方法,包括:包括在至少两个计算机处理器的每一个上提供该服务的逻辑;接收对该服务的请求报文,该报文具有IP地址,且用于将请求分配到具有提供该服务的逻辑的至少两个计算机处理器中的一个。 57. The IP address provided by the service addressing method, comprising: providing the service logic at each of the at least two computer processor; receiving the service request packet, the packet having an IP address, and It is assigned to a request for the at least two computer processor having a logic that provides the service in.
58.如权利要求57的方法,其特征在于分析输入报文的源信息,以确定哪个处理器将服务该报文。 58. The method of claim 57, wherein the source information analysis input packets to determine which processors will service the message.
CN 02811128 2001-04-20 2002-04-16 Virtual networking system and method in processing system CN1290008C (en)

Priority Applications (5)

Application Number Priority Date Filing Date Title
US28529601P true 2001-04-20 2001-04-20
US10/038,353 US7231430B2 (en) 2001-04-20 2002-01-04 Reconfigurable, virtual processing system, cluster, network and method
US10/038,355 US6971044B2 (en) 2001-04-20 2002-01-04 Service clusters and method in a processing system with failover capability
US10/038,354 US7174390B2 (en) 2001-04-20 2002-01-04 Address resolution protocol system and method in a virtual network
US10/037,191 US20030130832A1 (en) 2002-01-04 2002-01-04 Virtual networking system and method in a processing system

Publications (2)

Publication Number Publication Date
CN1520550A CN1520550A (en) 2004-08-11
CN1290008C true CN1290008C (en) 2006-12-13

Family

ID=27534612

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 02811128 CN1290008C (en) 2001-04-20 2002-04-16 Virtual networking system and method in processing system

Country Status (6)

Country Link
EP (1) EP1388057A4 (en)
JP (1) JP2005506726A (en)
CN (1) CN1290008C (en)
CA (1) CA2444066A1 (en)
DE (1) DE10296675T5 (en)
WO (1) WO2002086712A1 (en)

Families Citing this family (45)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7231430B2 (en) * 2001-04-20 2007-06-12 Egenera, Inc. Reconfigurable, virtual processing system, cluster, network and method
WO2003067480A1 (en) 2002-02-07 2003-08-14 Thinkdynamics Inc. Method and system for managing resources in a data center
US8122453B2 (en) 2003-02-04 2012-02-21 International Business Machines Corporation Method and system for managing resources in a data center
GB2406121B (en) * 2002-05-24 2005-12-28 Baker Hughes Inc A method and apparatus for high speed data dumping and communication for a down hole tool
US7457300B2 (en) 2003-01-21 2008-11-25 Telefonaktiebolaget L M Ericsson (Publ) Ethernet address management system
US7843907B1 (en) 2004-02-13 2010-11-30 Habanero Holdings, Inc. Storage gateway target for fabric-backplane enterprise servers
US7843906B1 (en) 2004-02-13 2010-11-30 Habanero Holdings, Inc. Storage gateway initiator for fabric-backplane enterprise servers
US7664110B1 (en) 2004-02-07 2010-02-16 Habanero Holdings, Inc. Input/output controller for coupling the processor-memory complex to the fabric in fabric-backplane interprise servers
US7990994B1 (en) 2004-02-13 2011-08-02 Habanero Holdings, Inc. Storage gateway provisioning and configuring
US7860097B1 (en) 2004-02-13 2010-12-28 Habanero Holdings, Inc. Fabric-backplane enterprise servers with VNICs and VLANs
US7685281B1 (en) 2004-02-13 2010-03-23 Habanero Holdings, Inc. Programmatic instantiation, provisioning and management of fabric-backplane enterprise servers
US7561571B1 (en) 2004-02-13 2009-07-14 Habanero Holdings, Inc. Fabric address and sub-address resolution in fabric-backplane enterprise servers
US8145785B1 (en) 2004-02-13 2012-03-27 Habanero Holdings, Inc. Unused resource recognition in real time for provisioning and management of fabric-backplane enterprise servers
US7953903B1 (en) 2004-02-13 2011-05-31 Habanero Holdings, Inc. Real time detection of changed resources for provisioning and management of fabric-backplane enterprise servers
US7757033B1 (en) 2004-02-13 2010-07-13 Habanero Holdings, Inc. Data exchanges among SMP physical partitions and I/O interfaces enterprise servers
US8868790B2 (en) 2004-02-13 2014-10-21 Oracle International Corporation Processor-memory module performance acceleration in fabric-backplane enterprise servers
US7633955B1 (en) 2004-02-13 2009-12-15 Habanero Holdings, Inc. SCSI transport for fabric-backplane enterprise servers
US7860961B1 (en) 2004-02-13 2010-12-28 Habanero Holdings, Inc. Real time notice of new resources for provisioning and management of fabric-backplane enterprise servers
US7873693B1 (en) 2004-02-13 2011-01-18 Habanero Holdings, Inc. Multi-chassis fabric-backplane enterprise servers
US7296092B2 (en) * 2004-06-30 2007-11-13 Nokia, Inc. Apparatus for inter-domain communications including a virtual switch for routing data packets between virtual interfaces of the virtual switch
CN100407823C (en) 2004-11-10 2008-07-30 中兴通讯股份有限公司 A communication system initialization and internal node addressing method
US8086755B2 (en) * 2004-11-29 2011-12-27 Egenera, Inc. Distributed multicast system and method in a network
JP4462024B2 (en) 2004-12-09 2010-05-12 株式会社日立製作所 Failover method according to the disk takeover
US8254285B2 (en) 2005-02-25 2012-08-28 Ip Infusion, Inc. Hardware abstraction layer
US7586936B2 (en) * 2005-04-01 2009-09-08 International Business Machines Corporation Host Ethernet adapter for networking offload in server environment
CN100393060C (en) 2005-04-01 2008-06-04 中兴通讯股份有限公司 Data communication method for device interior
KR100528171B1 (en) * 2005-04-06 2005-11-04 스콥정보통신 주식회사 Ip management method and apparatus for protecting/blocking specific ip address or specific device on network
CN100450080C (en) 2005-05-17 2009-01-07 华为技术有限公司 Method and apparatus for astringing two layer MAC address
JP4701929B2 (en) 2005-09-02 2011-06-15 株式会社日立製作所 Boot configuration changing, the management server and the computer system,
US8509218B2 (en) 2006-02-28 2013-08-13 Genband Us Llc Prioritization within a session over internet protocol (SOIP) network
US7564850B2 (en) 2007-01-17 2009-07-21 Huawei Technologies Co., Ltd. Method for transmitting layer 2 packet and access device thereof
US7912062B2 (en) * 2007-09-28 2011-03-22 Genband Us Llc Methods and apparatus for managing addresses related to virtual partitions of a session exchange device
JP4488077B2 (en) 2008-02-29 2010-06-23 日本電気株式会社 Virtualization system, virtualization method, and virtualization program
US8817656B2 (en) * 2008-04-08 2014-08-26 Cisco Technology, Inc. Discovery for fibre channel over ethernet devices
US9106540B2 (en) * 2009-03-30 2015-08-11 Amazon Technologies, Inc. Providing logical networking functionality for managed computer networks
CN101616042B (en) 2009-08-04 2012-09-05 中兴通讯股份有限公司 Network equipment simulator and method for sending, receiving and transmitting data
CN102118259B (en) * 2009-12-30 2013-06-05 华为技术有限公司 Method and device for creating Ethernet route
JP5316616B2 (en) * 2011-09-16 2013-10-16 株式会社日立製作所 Business takeover methods, computer systems, and management server
US9231846B2 (en) * 2011-11-22 2016-01-05 Microsoft Technology Licensing, Llc Providing network capability over a converged interconnect fabric
CN102624597B (en) * 2012-03-09 2014-12-17 北京邮电大学 Two-way sequencing virtual network mapping method
US20140040477A1 (en) * 2012-07-31 2014-02-06 F5 Networks, Inc. Connection mesh in mirroring asymmetric clustered multiprocessor systems
US9417831B2 (en) 2014-03-05 2016-08-16 Tricerat Method and system of providing computer network based limited visibility service discovery
WO2017051702A1 (en) * 2015-09-25 2017-03-30 株式会社東芝 Control device
CN106559322A (en) * 2015-09-25 2017-04-05 北京计算机技术及应用研究所 Security protection gateway based on multi-godson parallel processing framework
EP3300331A4 (en) * 2016-07-14 2018-05-16 Huawei Technologies Co., Ltd. Response method, apparatus and system in virtual network computing authentication, and proxy server

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH03148940A (en) * 1989-11-06 1991-06-25 Hitachi Ltd Mutual connection system for lan and isdn
CA2144743C (en) * 1993-07-28 2003-10-21 Jeffrey Krause Network station with multiple network addresses
GB9401092D0 (en) * 1994-01-21 1994-03-16 Newbridge Networks Corp A network management system
US5473599A (en) * 1994-04-22 1995-12-05 Cisco Systems, Incorporated Standby router protocol
US5805805A (en) * 1995-08-04 1998-09-08 At&T Corp. Symmetric method and apparatus for interconnecting emulated lans
US5684800A (en) * 1995-11-15 1997-11-04 Cabletron Systems, Inc. Method for establishing restricted broadcast groups in a switched network
JP2974280B2 (en) * 1996-09-11 1999-11-10 日本電気通信システム株式会社 Virtual group information management method in the bridge device of a network connection
US5835725A (en) * 1996-10-21 1998-11-10 Cisco Technology, Inc. Dynamic address assignment and resolution technique
US5970066A (en) * 1996-12-12 1999-10-19 Paradyne Corporation Virtual ethernet interface
US6091732A (en) * 1997-11-20 2000-07-18 Cisco Systems, Inc. Method for configuring distributed internet protocol gateways with lan emulation
US6178171B1 (en) * 1997-11-24 2001-01-23 International Business Machines Corporation Route switching mechanisms for source-routed ATM networks
US6195705B1 (en) * 1998-06-30 2001-02-27 Cisco Technology, Inc. Mobile IP mobility agent standby protocol

Also Published As

Publication number Publication date
CN1520550A (en) 2004-08-11
JP2005506726A (en) 2005-03-03
WO2002086712A1 (en) 2002-10-31
DE10296675T5 (en) 2004-04-22
EP1388057A4 (en) 2009-07-01
CA2444066A1 (en) 2002-10-31
EP1388057A1 (en) 2004-02-11

Similar Documents

Publication Publication Date Title
US7769004B2 (en) Network abstraction and isolation layer for masquerading machine identity of a computer
CN102668473B (en) For high-performance, low power systems and methods for data center interconnect structure
Patel et al. Ananta: Cloud scale load balancing
US7515589B2 (en) Method and apparatus for providing network virtualization
CN101156408B (en) Network communications for operating system partitions
CN102652305B (en) Virtual storage target unload technology
US7996560B2 (en) Managing virtual ports in an information processing system
EP1682978B1 (en) Method and system for accessing and managing virtual machines
US6249814B1 (en) Method and apparatus for identifying devices on a network
US6061349A (en) System and method for implementing multiple IP addresses on multiple ports
US8108454B2 (en) Address assignment in Fibre Channel over Ethernet environments
US6049528A (en) Trunking ethernet-compatible networks
US5923854A (en) Virtual internet protocol (IP) addressing
CN101601232B (en) Triple-tier anycast addressing
US7877512B2 (en) Isolation switch for fibre channel fabrics in storage area networks
US7363347B2 (en) Method and system for reestablishing connection information on a switch connected to plural servers in a computer network
US8677023B2 (en) High availability and I/O aggregation for server environments
CN102105865B (en) Method and system for power management in a virtual machine environment without disrupting network connectivity
EP2617165B1 (en) System and method for providing ethernet over infiniband virtual hub scalability in a middleware machine environment
US6490632B1 (en) High performance load balancing and fail over support of internet protocol exchange traffic over multiple network interface cards
JP4937902B2 (en) Management of the process state information in the operating system environment
US7836332B2 (en) Method and apparatus for managing virtual ports on storage systems
JP3945297B2 (en) System and management system
CN103793359B (en) A method for communication and a system virtual port
US7493409B2 (en) Apparatus, system and method for implementing a generalized queue pair in a system area network

Legal Events

Date Code Title Description
C06 Publication
C10 Request of examination as to substance
C14 Granted
C17 Cessation of patent right