CN102622304A - Memory optimizing method for double-layer address space mapping - Google Patents

Memory optimizing method for double-layer address space mapping Download PDF

Info

Publication number
CN102622304A
CN102622304A CN 201210029099 CN201210029099A CN102622304A CN 102622304 A CN102622304 A CN 102622304A CN 201210029099 CN201210029099 CN 201210029099 CN 201210029099 A CN201210029099 A CN 201210029099A CN 102622304 A CN102622304 A CN 102622304A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
address space
memory
mapping
physical
extended
Prior art date
Application number
CN 201210029099
Other languages
Chinese (zh)
Inventor
卢林发
叶灿才
黄家祺
Original Assignee
中山爱科数字科技股份有限公司
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

Links

Abstract

The invention discloses a memory optimizing method for double-layer address space mapping, which implements dynamic distribution of memory resources for cloud computing environment and cross resource boundaries of different physical servers. The memory optimizing method for double-layer address space mapping is characterized in that each physical server utilizes a double-layer address space mapping mechanism; a console of each physical server provides overall extended address space and manages all remote idle memories mapping to a local extended address space; a virtual machine monitor of each physical server provides a logical extended address space to expand a local physical address space, and the consoles of the physical servers provide address mapping components and complete mapping from the overall extended address spaces to the logical extended address space.

Description

—种双层地址空间映射的内存优化方法 - Memory Optimization kinds of double address space map

技术领域 FIELD

[0001] 本发明涉及云计算技木,特别是面向云计算环境的物理服务器内存资源动态分配和利用的方法。 [0001] The technology of wood, in particular the present invention relates to a method for dynamic allocation and cloud computing utilizing physical server memory resources for cloud environment.

背景技术 Background technique

[0002] 云计算使计算机成为了ー种效能工具、使软件作为ー种服务,同时对计算机软件和硬件设计都带来了巨大的挑战。 [0002] Cloud computing allows computer became ー kinds of performance tools that enable software-as-ー kinds of services, and computer software and hardware designs have brought a great challenge. 对于支撑服务运行的底层计算基础设施,挑战主要集中在两个方面:系统软件和硬件。 Support for the underlying computing infrastructure services running, the challenge focused on two areas: system software and hardware. ,自适应性资源配置是系统软件面临的最大挑战,如何分配资源又是这个挑战的最核心问题。 , Adaptive allocation of resources is the biggest challenge facing the system software, how to allocate resources is the core issue the challenge.

[0003] 虚拟化技术的引入带来了如下几个方面的好处:更细粒度的资源共享、安全隔离、高可靠性等。 [0003] introduction of virtualization technology brings the benefits of the following aspects: the more fine-grained resource sharing, security isolation, high reliability. 通过以虚拟机(VM)为粒度的分时复用技术,虚拟化平台可以有效地进行服务器整合以提升资源利用率,并最终提高云计算基础设施的整体资源利用率。 By time-multiplexing virtual machine (VM) for the size of the virtual platform can effectively consolidate servers to improve resource utilization, and ultimately improve the overall cloud infrastructure resource utilization. .但是,现在的虚拟化技术主要关注在一台物理服务器内的资源优化,通过资源的复用技术来提高单台服务器的资源利用率,而对于全局资源优化来说,则缺乏访问远程资源的虚拟化支撑技木。 But now we focus on virtualization technology to optimize resources within a single physical server, by multiplexing resources to improve resource utilization single server, and for global resource optimization, it is the lack of access to remote resources support virtualization technology of wood.

[0004] 利用虚拟化技术来构建对操作系统透明的云数据中心面临如下两大挑战:(I)如何构建更有弹性的、可靠的全局资源空间,并建立资源流动框架,实现跨越物理机边界的资源优化;(2)如何动态地识别闲置资源,为资源优化提供可以流动的资源空间。 [0004] to build the operating system transparent cloud data centers face two major challenges as the use of virtualization technology: (I) How to build a more resilient and reliable global resource space, resource flows and establish the framework to achieve across physical machine boundaries resource optimization; (2) how to dynamically identify idle resources, spatial resources for resource optimization can flow.

发明内容 SUMMARY

[0005] 本发明的目的是面向云计算环境,提出一种跨越不同物理服务器资源边界实现内存资源动态分配方案。 Objective [0005] The present invention is for a cloud computing environment, across different physical servers to provide a dynamic resource allocation scheme boundary memory resources. 它适合不需要修改操作系统源代码或插入对应的驱动来利用远程资源。 It is suitable not need to modify the operating system source code is inserted a corresponding drive or remote resources utilized.

[0006] 本发明通过如下方案实现: [0006] The present invention is achieved by the following scheme:

[0007] —种双层地址空间映射的内存优化方法,为面向云计算环境,跨越不同物理服务器资源边界实现内存资源动态分配,其特征在于: [0007] - Optimization bilayer seed memory address space mapping for the cloud computing environment, the server resources across different physical boundaries of dynamic allocation of memory resources, comprising:

[0008] 各物理服务器采用双层地址空间映射机制;物理服务器的控制台提供全局扩展地址空间,管理所有映射到本地扩展地址空间的远程空闲内存;物理服务器的虚拟机监控器提供逻辑扩展地址空间,扩展本地物理地址空间;物理服务器的控制台提供地址映射组件,完成从全局扩展地址空间到逻辑扩展地址空间的映射。 [0008] Each physical address space mapping server using double mechanism; physical server console provides extended global address space, all mapped to the local management of the extended address space remote free memory; physical server virtual machine monitor providing the logical address space extension , extended local physical address space; physical server console provides address mapping assembly, the mapping from the global address space to the extended address space is a logical extension.

[0009] 其中,所述控制台为具有特权的虚拟机,拥有管理远程空闲内存的权限。 [0009] wherein the virtual machine console privileged, you have permission to manage remote free memory.

[0010] 进ー步,所述的双层地址空间映射的内存优化方法,其特征在于,包括步骤: [0010] Step ー into the memory address space mapping optimization method the bilayer, which is characterized by comprising the steps of:

[0011] 步骤一,各物理服务器虚拟机将本地空闲的内存资源通过虚拟技术形成远程空闲内存池; [0011] Step 1. Each physical server virtual machine resources local spare memory cell is formed by a remote virtual idle memory technology;

[0012] 步骤ニ,远程空闲内存池内的内存资源分别被映射到各物理服务器的本地扩展地址空间; [0012] Step ni, remote free memory pool of memory resources are mapped to the local address space of each expansion physical server;

[0013] 步骤三,各物理服务器通过虚拟机监控器监控远程可用的内存;[0014] 步骤四,各物理服务器通过控制台的全局扩展地址空间管理所有映射到本地扩展地址空间的远程空闲内存;并通过地址映射组件完成从全局扩展地址空间到逻辑扩展地址空间的映射; [0013] Step three, each physical server through the virtual machine monitor to monitor remotely available memory; [0014] Step 4. Each mapped to the physical remote servers manage all the free space of the local memory expansion extended address by the global address space of the console; and the mapping from the global address space to extend the logical address space is extended by address mapping assembly;

[0015] 步骤五,各物理服务器通过虚拟机监控器的逻辑扩展地址空间扩展本地物理地址空间。 [0015] Step 5. Each physical server extended local physical address space by a logical extension of the address space of the virtual machine monitor.

[0016] 各物理服务器的空闲内存通过服务器ID、地址段来表示其地址构成。 [0016] Idle memory of each physical server through the server ID, the address segment configured to indicate its address.

[0017] 综上所述,本发明方法具有如下显著特点和进步: [0017] In summary, the method of the present invention has the following significant features and progress:

[0018] 1)、采用双层地址空间映射机制,方便虚拟内存资源和实体内存资源映射和管理。 [0018] 1), using double-address space mapping mechanism to facilitate virtual memory and physical memory resource management and resource mapping.

[0019] 2)、能有效解决跨越不同物理服务器资源边界实现内存资源动态分配; [0019] 2), can effectively solve the physical server resources across different memory boundaries to achieve dynamic resource allocation;

[0020] 3)、不需要修改操作系统源代码或插入对应的驱动来利用远程内存资源。 [0020] 3), without modifying the operating system source code to or inserted into the corresponding drive using a remote memory resources.

附图说明 BRIEF DESCRIPTION

[0021] 图I是空闲内存池架构图; [0021] FIG. I is a schematic diagram of the free memory pool;

[0022] 图2是双层空间映射图。 [0022] FIG. 2 is a double maps.

具体实施方式 detailed description

[0023] 参考图1,云计算环境中,大量的物理服务器组成服务器集群。 [0023] Referring to FIG 1, a cloud computing environment, a large number of physical servers server cluster. 物理服务器上运行有一个或一个以上的虚拟机。 Run one or more virtual machines on a physical server. 虚拟机由于其负载的动态变化,其内存利用率在一定的时间内会随负载变化产生波动,所以,每台物理服务器内部及不同物理机之间内存的利用率就呈现不均衡性,并且这种不均衡性随时间而动态变化。 Due to the dynamic changes in the virtual machine of its load, memory utilization it will produce within a certain period of time varies with load fluctuation, so the memory of each physical server utilization between different physical machines and internal imbalance on the show, and this kind of imbalance and dynamic changes over time. 低负载压力的虚拟机释放空闲内存进入空闲内存池,高负载压力的虚拟机从空闲内存池中申请内存来降低其内存压力。 Low load pressure of the virtual machine is released into the free memory free memory pool, high-load pressure from the application of virtual machine memory pool of free memory to reduce the memory pressure.

[0024] 参考图2,本发明物理服务器采用双层地址空间映射架构。 [0024] Referring to FIG 2, the present invention employs a double physical server address space mapping schema. 其中,控制台,即拥有特权的虚拟机,它拥有管理远程空闲内存的权限。 Among them, console, virtual machine that is privileged, it has the authority to manage remote free memory. 在控制台内分别设全局扩展地址空间和地址映射组件。 Respectively provided in the console and the global address of the extended address space mapping component. 虚拟机监控器内具有逻辑扩展地址空间。 It has a logical address space within the extended virtual machine monitor. 远程空闲内存池的内存资源经过地址映射组件映射和逻辑扩展地址空间扩展成为物理服务器本地的内存资源。 Remote free memory pool of memory resources through the address mapping component mapping and logical extension of the address space expanded into a physical server's local memory resources. 全局扩展地址空间作用是管理所有映射到本地扩展地址空间的远程空闲内存;而逻辑扩展地址空间用于扩展本地物理地址空间;地址映射组件则从全局扩展地址空间到逻辑扩展地址空间的映射。 Extended global role is to manage all address space mapped to the local address space of the extended remote free memory; and a logical address space for extended extended local physical address space; assembly from the global address mapping to extended mapping the logical address space of the extended address space. 逻辑扩展地址空间由一个从零开始的独立地址空间和一个标识其空间的全局唯一标识符构成。 A globally unique identifier of a logical zero-extended address space independent address space identifier and a spatial configuration. 通过(地址,标识符)对来对空间进行寻址,并在地址映射组件的翻译下最终确定要访问的远程内存页面.逻辑扩展地址空间的申请和回收都是由虚拟机监控器来控制,这样可以提供更为弹性的空间使用模式。 To be addressed by space (address, identifier), and ultimately determines to access remote memory pages. Application address space and recovering the logical extension is controlled by a virtual machine monitor in the translation address mapping assembly, this provides a more flexible space usage patterns. 地址映射组件的核心功能是提供全局扩展地址空间和逻辑扩展地址空间之间的双向映射,通过维护相关的映射关系来提供高效可靠的远程内存存储抽象。 Address mapping component of the core function is to provide a global expansion of two-way mapping between the logical address space and extended address space, to provide efficient and reliable remote memory storage abstraction by maintaining relevant mappings. 地址映射组件支持两类基本的数据操作:数据存取和数据同步。 Address mapping component supports two basic operations of data: data access and data synchronization. 其中,数据存取是基本的远程内存操作,实现对于远程数据的读取和写入.数据同步操作为远程内存服务的退出提供支持,通过从退出服务器到其它备份服务器同步数据可以实现数据存储的高可靠性。 Wherein the data access operation is substantially remote memory, to achieve reading and writing of data to the remote data synchronization exit remote memory support and services, data storage can be achieved by synchronizing data from the server to the other exit backup server high reliability.

[0025] 进一步,本发明实施包括如下步骤: [0025] Further, embodiments of the present invention comprises the steps of:

[0026] 步骤一,各物理服务器虚拟机将本地空闲的内存资源通过虚拟技术形成远程空闲内存池;[0027] 步骤ニ,远程空闲内存池内的内存资源分别被映射到各物理服务器的本地扩展地址空间; [0026] Step 1. Each physical server virtual machine local free memory resources forming the remote free memory pool by the virtual technology; [0027] Step ni, remote free memory pool of memory resources are mapped to the local extension addresses for each physical server space;

[0028] 步骤三,各物理服务器通过虚拟机监控器监控远程可用的内存; [0028] Step three, each of the physical memory available remote monitoring server via a virtual machine monitor;

[0029] 步骤四,各物理服务器通过控制台的全局扩展地址空间管理所有映射到本地扩展地址空间的远程空闲内存;并通过地址映射组件完成从全局扩展地址空间到逻辑扩展地址空间的映射;各物理服务器的操作系统通过接ロ操作全局扩展地址空间时包括如下步骤:添加一段远程内存到全局扩展地址空间,返回该内存段在地址空间中的起始地址;删除一段全局地址空间;影射一段全局地址空间到逻辑扩展地址空间,返回其在逻辑扩展地址空间中的起始地址;释放从逻辑扩展地址空间中已经扩展为本地内存的空间映射。 [0029] Step 4. Each physical server by mapping all of the global address space management console is extended to the extended address space remote local free memory; and the mapping from the global address space to the extended address space by logical extension address mapping assembly; each operating system server includes physical contact by the operator ro extended global address space the steps of: adding some remote memory address space to a global expansion, the return to the starting address of the memory segment in the address space; for removal of a global address space; global mapping section the logical address space to the extended address space, which returns the starting address in a logical address space in extended; extended release logical address space has been expanded to the local memory space mapping. [0030] 步骤五,各物理服务器通过虚拟机监控器的逻辑扩展地址空间扩展本地物理地址空间。 [0030] Step 5. Each physical server extended local physical address space by a logical extension of the address space of the virtual machine monitor.

[0031] 另外,本发明所提到的控制台、虚拟机、虚拟机监控器等可以參考现有软件技术编程实现。 [0031] Further, the present invention is referred to the console, the virtual machine, the virtual machine monitor or the like may refer to the prior art of software programming. 在不脱离本权利要求书记载技术方案范围的等效变换应属于本发明保护范围内。 In the claims without departing from the scope of the technical solution according to equivalent transformation should fall within the scope of the invention.

Claims (5)

  1. 1. 一种双层地址空间映射的内存优化方法,为面向云计算环境,跨越不同物理服务器资源边界实现内存资源动态分配,其特征在于: 各物理服务器采用双层地址空间映射机制;物理服务器的控制台提供全局扩展地址空间,管理所有映射到本地扩展地址空间的远程空闲内存;物理服务器的虚拟机监控器提供逻辑扩展地址空间,扩展本地物理地址空间;物理服务器的控制台提供地址映射组件,完成从全局扩展地址空间到逻辑扩展地址空间的映射。 1. A method of optimizing memory address space mapping bilayer, cloud computing environment for, the boundary across different physical server resources to achieve dynamic allocation of memory resources, wherein: each of the physical address space mapping server using double mechanism; physical server the console provides global extended address space, all mapped to the local administration extended address space of the remote free memory; physical server virtual machine monitor provides a logical extension of the address space, expand the local physical address space; physical server's console provides address mapping component, the mapping from the global address space to the extended address space is a logical extension.
  2. 2.如权利要求I所述的双层地址空间映射的内存优化方法,其特征在于,所述控制台为具有特权的虚拟机,拥有管理远程空闲内存的权限。 2. I claim the memory address space mapping optimization method the bilayer, wherein the virtual machine console with privileges, rights management have remote free memory.
  3. 3.如权利要求2所述的双层地址空间映射的内存优化方法,其特征在于,包括步骤: 步骤一,各物理服务器虚拟机将本地空闲的内存资源通过虚拟技术形成远程空闲内存池; 步骤二,远程空闲内存池内的内存资源分别被映射到各物理服务器的本地扩展地址空间; 步骤三,各物理服务器通过虚拟机监控器监控远程可用的内存; 步骤四,各物理服务器通过控制台的全局扩展地址空间管理所有映射到本地扩展地址空间的远程空闲内存;并通过地址映射组件完成从全局扩展地址空间到逻辑扩展地址空间的映射; 步骤五,各物理服务器通过虚拟机监控器的逻辑扩展地址空间扩展本地物理地址空间。 Memory address space mapping optimization bunk as claimed in claim 2, further comprising the step of: step 1, each local server virtual machine physical memory resources forming the free memory pool by the virtual remote free technology; step Second, remote free memory pool of memory resources are mapped to each physical server's local extension address space; step three, each of the physical memory available remote monitoring server via a virtual machine monitor; step four, each physical server console by the global extended address space management all mapped to the local extension address space remote free memory; and the mapping from the global extended address space to the logical extension of the address space by address mapping assembly; step five, each physical server logic extended address by a virtual machine monitor spatial expansion of local physical address space.
  4. 4.如权利要求3所述的双层地址空间映射的内存优化方法,其特征在于,各物理服务器的空闲内存通过服务器ID、地址段来表示其地址构成。 Memory address space mapping optimization of bilayer as claimed in claim 3, wherein each physical memory of the server is idle through the server ID, the address segment configured to indicate its address.
  5. 5.如权利要求4所述的双层地址空间映射的内存优化方法,其特征在于,各物理服务器的操作系统通过接口操作全局扩展地址空间时包括如下步骤: 添加一段远程内存到全局扩展地址空间,返回该内存段在地址空间中的起始地址; 删除一段全局地址空间; 影射一段全局地址空间到逻辑扩展地址空间,返回其在逻辑扩展地址空间中的起始地址; 释放从逻辑扩展地址空间中已经扩展为本地内存的空间映射。 Memory address space mapping optimization of bilayer as claimed in claim 4, wherein the operating system includes expansion of each physical server address space by operating the global interface the steps of: adding some remote memory address space to a global expansion , return to the starting address of the memory address space segment; delete section of a global address space; global address space mapping for some logical extension to the address space, which returns the starting address in a logical address space in extended; extended release of the logical address space It has been extended to local memory space mapping.
CN 201210029099 2012-02-07 2012-02-07 Memory optimizing method for double-layer address space mapping CN102622304A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201210029099 CN102622304A (en) 2012-02-07 2012-02-07 Memory optimizing method for double-layer address space mapping

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201210029099 CN102622304A (en) 2012-02-07 2012-02-07 Memory optimizing method for double-layer address space mapping

Publications (1)

Publication Number Publication Date
CN102622304A true true CN102622304A (en) 2012-08-01

Family

ID=46562231

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201210029099 CN102622304A (en) 2012-02-07 2012-02-07 Memory optimizing method for double-layer address space mapping

Country Status (1)

Country Link
CN (1) CN102622304A (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101398768A (en) * 2008-10-28 2009-04-01 北京航空航天大学 Construct method of distributed virtual machine monitor system
CN101470633A (en) * 2007-12-24 2009-07-01 联想(北京)有限公司 Virtual machine monitor, virtual machine system and its internal memory processing method
CN101969391A (en) * 2010-10-27 2011-02-09 北京邮电大学 Cloud platform supporting fusion network service and operating method thereof
US20110082997A1 (en) * 2009-10-04 2011-04-07 Infinidat Ltd. Virtualized storage system and method of operating thereof
CN102193824A (en) * 2010-03-18 2011-09-21 微软公司 Virtual machine homogenization to enable migration across heterogeneous computers
US20110296234A1 (en) * 2010-05-25 2011-12-01 Microsoft Corporation Virtual machine i/o multipath configuration

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101470633A (en) * 2007-12-24 2009-07-01 联想(北京)有限公司 Virtual machine monitor, virtual machine system and its internal memory processing method
CN101398768A (en) * 2008-10-28 2009-04-01 北京航空航天大学 Construct method of distributed virtual machine monitor system
US20110082997A1 (en) * 2009-10-04 2011-04-07 Infinidat Ltd. Virtualized storage system and method of operating thereof
CN102193824A (en) * 2010-03-18 2011-09-21 微软公司 Virtual machine homogenization to enable migration across heterogeneous computers
US20110296234A1 (en) * 2010-05-25 2011-12-01 Microsoft Corporation Virtual machine i/o multipath configuration
CN101969391A (en) * 2010-10-27 2011-02-09 北京邮电大学 Cloud platform supporting fusion network service and operating method thereof

Similar Documents

Publication Publication Date Title
US20100235580A1 (en) Multi-Domain Management of a Cache in a Processor System
US20100094948A1 (en) Workload migration using on demand remote paging
US20120079232A1 (en) Apparatus, method, and system for implementing micro page tables
Berezecki et al. Many-core key-value store
US20060020769A1 (en) Allocating resources to partitions in a partitionable computer
US20110320682A1 (en) Cooperative memory resource management via application-level balloon
Hwang et al. Designing SSI clusters with hierarchical checkpointing and single I/O space
Marty et al. Virtual hierarchies to support server consolidation
US20090153897A1 (en) Method, System and Program Product for Reserving a Global Address Space
CN101398768A (en) Construct method of distributed virtual machine monitor system
CN101997918A (en) Method for allocating mass storage resources according to needs in heterogeneous SAN (Storage Area Network) environment
US20140123145A1 (en) Efficient memory virtualization in multi-threaded processing units
CN101847105A (en) Computer and internal memory sharing method of a plurality of operation systems
US8812400B2 (en) Managing a memory segment using a memory virtual appliance
US20090157996A1 (en) Method, System and Program Product for Allocating a Global Shared Memory
US20120030406A1 (en) Hypervisor-based management of local and remote virtual memory pages
CN102193814A (en) Method and system for dynamically distributing embedded virtual memory
US20090228635A1 (en) Memory Compression Implementation Using Non-Volatile Memory in a Multi-Node Server System With Directly Attached Processor Memory
US20110173622A1 (en) System and method for dynamic task migration on multiprocessor system
US20120047313A1 (en) Hierarchical memory management in virtualized systems for non-volatile memory models
Islam et al. Triple-H: A hybrid approach to accelerate HDFS on HPC clusters with heterogeneous storage architecture
Song et al. A case for scaling applications to many-core with OS clustering
US20130275703A1 (en) Switching optically connected memory
CN102523258A (en) Data storage framework facing cloud operation system and load balancing method thereof
Chiu et al. Evaluating caching and storage options on the Amazon Web Services Cloud

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C02 Deemed withdrawal of patent application after publication (patent law 2001)