JP2011203810A - Server, computer system, and virtual computer management method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To normally operate service on a virtual computer by taking into consideration a performance difference of hardware resources constituting a computer system, in a virtual computer monitor for managing the virtual computer on a physical computer.SOLUTION: A server includes a virtual machine monitor for managing the virtual computer and an operating system for executing an application so as to provide a virtual environment. The virtual machine monitor includes a physical resource information manager for managing information related to physical resources assigned to each virtual computer, an application information manager for managing information related to the operating state of the application, and a performance difference information manager for obtaining performance difference information representing a performance difference between the first physical resource performance and the second physical resource performance and generating control information to adjust the second physical resource performance so that the application is operated normally as long as a first physical resource in the server is changed to a second physical resource having higher performance.

Description

  The present invention relates to a virtual machine, and more particularly to a server, a computer system, and a virtual machine management method for transferring data using a virtual network interface that connects the virtual machine and another device.

  In recent years, a plurality of multi-core processors have been installed in physical computers, and server virtualization technology that simultaneously operates a plurality of virtual computers has attracted attention in order to effectively use processor resources.

  By using the server virtualization technology, a plurality of virtual computers can be operated on one physical computer, so that a cost reduction can be expected as compared with a conventional system in which one physical computer executes one OS.

  An open system such as a PC has a rapid progress in performance improvement, and successor models have appeared one after another, and the product warranty period is as short as about 5 years.

  In contrast, services such as core systems (business applications such as corporate accounting, production management, human resources, salary, and services related to social infrastructure such as banks and railways) are basic services. There are many things that do not change. In addition, the core service must not cause trouble by renewing a service that has been used. For the reasons described above, there is a need to operate the service as long as possible.

  The virtualization technology answers the above-mentioned needs. Even when new hardware is installed in a physical computer, it is possible to migrate the service environment that is currently in operation, and open systems Application to services that are cost-effective for the system has begun.

  When operating a service that you want to operate for a long time on a virtual machine, it is important to maintain the performance of the service. The hardware needs to be changed in a few years, and the performance difference increases with each change. Therefore, in order to maintain the environment when the service is built, the virtual machine monitor that controls the virtual machine must eliminate the performance difference and the environment in which the service is executed will not be affected by hardware changes. is required.

  One of the hardware resources is a network. As for the network, high speed and integration of a LAN (Local Area Network) network and a storage network are progressing.

  Ethernet (registered trademark, hereinafter the same) has a performance of 10 Gbps or more, and in particular, a data center (hereinafter referred to as DC) has many services that handle a large amount of data in real time. Is formulated by the IEEE 802.1 WG.

  A large amount of data can be handled at high speed on the network side. However, since the system configuration beyond the service's operating environment (virtual computer) exceeds the service reception capability, the service side can receive data on the virtual computer side. There is a possibility of exceeding the side processing. That is, an overflow of the reception buffer occurs, which causes data loss and abnormal termination of service or abnormal operation.

  As a method for solving this problem, there is a method for preventing a reception buffer of a destination virtual machine from overflowing in a virtual network system connecting a plurality of virtual machines generated on a certain physical machine (for example, Patent Document 1).

  Specifically, the virtual machine monitor includes a virtual network control unit that manages the transmission / reception status of the virtual machine, and each virtual machine includes a virtual network interface card (hereinafter referred to as a virtual NIC). The virtual network control unit monitors the reception buffer size of each virtual NIC, and determines whether or not transmission is possible from the transmission data size and reception buffer size transmitted by a certain virtual NIC. If the reception buffer remaining amount is less than or equal to the reception data size, transmission is stopped.

JP 2009-296133 A

  In Patent Document 1, since virtual machines are arranged on the same physical computer, the virtual machine monitor can grasp the data transmission / reception status of each virtual machine. Therefore, it is possible to compare the reception buffer of the reception-side virtual machine and the transmission data of the transmission-side virtual machine. However, since the physical NIC is not used, the relationship between the performance of the physical NIC and the operating environment of the virtual machine is not considered, and it is difficult to apply to a computer system that communicates with another physical computer via the physical NIC. It is.

  That is, the virtual machine monitor can manage virtual machines on the same physical machine, but cannot manage other machines, and therefore cannot be applied to a computer system including a plurality of machines.

  An object of the present invention is to appropriately consider the performance of hardware resources of a physical computer to be changed by a virtual computer monitor that controls a virtual computer generated on the physical computer in a computer system including a plurality of physical computers. It is to operate the application on the virtual machine normally.

  A typical example of the present invention is as follows. That is, a virtual machine monitor that includes a processor and a memory connected to the processor, and that is executed by the processor and manages a plurality of virtual machines generated by virtually dividing a physical resource of a server; A server that provides a virtual environment including an operating system that executes an application on the virtual machine, wherein the virtual machine monitor stores information on the physical resources allocated for each virtual calculation period Manages the physical resource information management unit to manage, the application information management unit that manages information related to the operating state of the application executed on the operating system for each virtual machine, and manages the physical resource performance information of the server And the first physical resource of the server is a second physical resource with high performance. When changed, it acquires performance difference information representing the difference between the performance of the first physical resource and the performance of the second physical resource, and the application operates normally based on the performance difference information And a performance difference information management unit for generating control information for adjusting the performance of the second physical resource.

  According to the present invention, when changing the hardware resources constituting the computer system, it is possible to operate normally without changing the operating state of the application executed on the virtual computer.

It is a block diagram explaining the outline | summary of embodiment of this invention. It is a block diagram explaining the structural example of the network system of the 1st Embodiment of this invention. It is a block diagram explaining an example of the internal structure of the server system of the 1st Embodiment of this invention. It is a block diagram explaining an example of the hardware constitutions of the physical computer of the 1st Embodiment of this invention. It is explanatory drawing which shows an example of the format of AP information management table of the 1st Embodiment of this invention. It is explanatory drawing which shows an example of the format of the structure log | history management table of the 1st Embodiment of this invention. It is a sequence diagram explaining the flow of the process which the server system in the network system of the 1st Embodiment of this invention performs. It is a flowchart explaining the process which the virtual machine monitor of the 1st Embodiment of this invention performs. It is a flowchart explaining the process which the virtual machine monitor of the 1st Embodiment of this invention performs. It is a flowchart explaining the logging process of AP operation information which the virtual machine monitor of the 1st Embodiment of this invention performs. It is a block diagram explaining an example of the internal structure of the network switch of the 2nd Embodiment of this invention. It is a sequence diagram explaining the flow of the process which the server system and network switch in the network system of the 2nd Embodiment of this invention perform. It is a flowchart explaining the process for eliminating the performance difference which the virtual machine monitor of the 2nd Embodiment of this invention performs. It is a flowchart explaining the process which the network switch of the 2nd Embodiment of this invention performs. It is a flowchart explaining the switching process performed by the network switch of the 2nd Embodiment of this invention. It is explanatory drawing which shows an example of the format of the AP-network environment information table which a network switch manages.

  Embodiments of the present invention will be described below with reference to the drawings. First, the outline of the present invention will be described with reference to FIG.

  FIG. 1 is a block diagram illustrating an outline of an embodiment of the present invention.

  In the example illustrated in FIG. 1, the physical server includes hardware 107 and software 100. The software 100 includes a virtual machine monitor 101 and a virtual machine 102.

  Hereinafter, a case where the hardware 107 is updated to the hardware 108 will be described as an example. Hereinafter, the hardware 107 will be referred to as old hardware 107 and the hardware 108 will be referred to as new hardware 108.

  The virtual machine monitor 101 includes application operation information 106, hardware performance difference information 109, and a performance difference elimination processing unit 110.

  The application operation information 106 stores information related to the operating environment of an application that operates on the virtual computer 102. The hardware performance difference information 109 stores information about the performance of the old hardware 107 and the new hardware 108. The performance difference elimination processing unit 110 executes processing for optimizing the difference in hardware performance when the hardware is changed.

  Specifically, the performance difference elimination processing unit 110 acquires the performance information of the new hardware 108 (S120), calculates the performance difference from the old hardware 107 (S121), and based on the changed hardware. The hardware performance difference to be optimized is set (S122).

[First Embodiment]
Next, a first embodiment of the present invention will be described with reference to FIGS.

  In the first embodiment, when the physical NIC is changed to a high-speed physical NIC in the physical computer, the virtual machine monitor that controls the virtual machine is based on the application operation information 106 and the hardware performance difference information 109. A pseudo congestion state is generated so that the reception buffer does not overflow.

  FIG. 2 is a block diagram illustrating a configuration example of the network system according to the first embodiment of this invention.

  The DC 200 includes a router 202, network switches 203-1 to 203-n, server systems 204-1 to 204-n, and storage apparatuses 205-1 and 205-2. Hereinafter, when the network switches 203-1 to 203-n are not distinguished from each other, they are described as the network switch 203. When the server systems 204-1 to 204-n are not distinguished from each other, they are referred to as the server system 204, and the storage device 205- 1 and 205-2 are described as the storage device 205 when not distinguished.

  The router 202, the network switch 203, the server system 204, and the storage apparatus 205 are connected via a network 206, respectively.

  The DC 200 is connected to another DC, system, terminal, or the like via an external network 201 such as the Internet.

  FIG. 3 is a block diagram illustrating an example of the internal configuration of the server system 204 according to the first embodiment of this invention.

  In the example illustrated in FIG. 3, the server system 204 includes a physical computer 300 and a physical NIC 312. The physical computer 300 and the physical NIC 312 are connected to each other via an internal bus 320 such as a PCI (Peripheral Component Interconnect) bus.

  The physical computer 300 includes one or more virtual machines 1 (301-1) to n (301-n) and a virtual machine monitor 302. The hardware configuration of the physical computer 300 will be described later with reference to FIG.

  The virtual machine monitor 302 generates virtual machine 1 (301-1) to virtual machine n (301-n), and manages the generated virtual machine 1 (301-1) to virtual machine n (301-n). .

  The virtual machine monitor 302 converts physical computer resources included in the physical machine 300 into virtual machine resources, and each of the virtual machine monitors 302 is transferred to the virtual machine 1 (301-1) to the virtual machine n (301-n). Allocate computer resources. Note that a method of allocating computer resources included in the physical computer 300 is not described in detail here because a known or publicly known technique may be used as appropriate.

  Hereinafter, when the virtual machines 1 (301-1) to n (301-n) are not distinguished, they are referred to as virtual machines 301.

  As a physical computer resource provided in the physical computer 300, the one shown in FIG. 4 is conceivable.

  FIG. 4 is a block diagram illustrating an example of the hardware configuration of the physical computer 300 according to the first embodiment of this invention.

  As shown in FIG. 4, the physical computer 300 includes a CPU 330, a memory 331, and a non-volatile storage medium 332, which are connected to each other via an internal bus or the like.

  The CPU 330 executes a program developed on the memory 331. The memory 331 stores a program executed by the CPU 330 and information necessary for executing the program. The memory 331 is also used as a work area for the CPU 330 to execute processing. The non-volatile storage medium 332 stores various types of information, such as an HDD.

  The virtual machine monitor 302 assigns the CPU 330, the memory 331, and the nonvolatile storage medium 332 to each virtual machine 301. Note that the physical computer 300 may have a configuration (not shown). For example, an I / O interface, a network interface, or the like may be provided.

  Note that FIG. 4 shows the case where there is one CPU 330, one memory 331, and one non-volatile storage medium 332, but a plurality of them may be provided. The CPU 330 may include a plurality of cores.

  Returning to the description of FIG.

  The virtual machine 1 (301-1) to the virtual machine n (301-n) are applications (hereinafter referred to as AP) 321-1 to 321-n, guest OS 322-1 to 322-n, and virtual driver 323-1. ~ 323-n.

  The guest OSs 322-1 to 322-n are OSs executed on the respective virtual machines 301. The APs 321-1 to 321-n are programs executed on the guest OSs 322-1 to 322-n. When the APs 321-1 to 321-n are executed, that is, when the APs 321-1 to 322-n operate, services corresponding to the APs 321-1 to 322-n are provided.

  Hereinafter, the guest OSs 322-1 to 322-n are described as guest OS 322 when not distinguished from each other, and the APs 321-1 to 321-n are described as AP 321 when not distinguished from each other.

  The virtual drivers 323-1 to 323-n are device drivers used when the guest OSs 322-1 to 322-n execute processing.

  Hereinafter, when the virtual drivers 323-1 to 323-n are not distinguished, they are described as virtual drivers 323.

  In this embodiment, network emulation processing in each virtual machine 301 is executed by the physical NIC 312.

  The virtual machine monitor 302 includes a transmission monitor 303, a reception monitor 304, a performance difference elimination processing unit 305, a pseudo congestion processing unit 306, an AP information management unit 307, an AP information management table 308, a configuration history management unit 309, and a configuration history management table. 310 is provided.

  The transmission monitor 303 analyzes the data transmitted by the AP 321 executed on each virtual machine 301 and transmits the analysis result to the physical NIC 312.

  The reception monitor 304 logs data addressed to each AP 321 executed on the virtual machine 301 and received via the physical NIC 312. For example, the reception interval of reception data, the number of reception data buffered in the reception buffer of the virtual NIC, and the like are monitored.

  The AP information management unit 307 manages information related to the operating environment of the AP 321. Hereinafter, information regarding the operating environment of the AP 321 is referred to as AP operation information.

  The AP operation information may be, for example, an identifier that identifies the AP 321, network setting information, a reception interval of reception data measured by the reception monitor 304, and the like.

  The AP information management unit 307 acquires the AP operation information as described above, and stores the acquired AP operation information in the AP information management table 308. Details of the AP information management table 308 will be described later with reference to FIG.

  The configuration history management unit 309 manages the hardware configuration information of the server system 204 and the hardware configuration information assigned to the virtual machine 301.

  The configuration history management unit 309 acquires, for example, the hardware configuration information of the server system 204 and the hardware configuration information assigned to the virtual computer 301 when the virtual computer 301 is generated, and the acquired configuration information is managed as a configuration history. Store in table 310. Note that the hardware configuration information of the server system 204 may be acquired when the server system 204 is initially set.

  When the hardware configuration of the server system 204 is changed, the configuration history management unit 309 acquires the changed hardware configuration information and stores the acquired hardware configuration information in the configuration history management table 310. .

  Details of the configuration history management table 310 will be described later with reference to FIG.

  The performance difference elimination processing unit 305 determines control information for operating the AP 321 normally when the hardware configuration of the server system 204 is changed.

  Specifically, when the hardware configuration of the server system 204 is changed, the performance difference elimination processing unit 305 includes the hardware configuration information at the time when the AP 321 is installed and the hardware after the hardware configuration is changed. Hardware configuration information is acquired from the configuration history management table 310. Further, the performance difference elimination processing unit 305 calculates a difference between the hardware configuration information at the time when the AP 321 is installed and the hardware configuration information after the hardware configuration is changed, and calculates the difference. Based on this, the control content for operating the AP 321 normally is determined. In the present embodiment, the hardware to be monitored is described as the physical NIC 312, but the present invention can be implemented even with hardware such as the CPU 330 or the memory 331.

  The pseudo congestion processing unit 306 executes a process for generating a pseudo congestion state based on the control content determined by the performance difference cancellation processing unit 305.

  The physical NIC 312 is a network interface for the server system 204 to communicate with other devices (for example, the network switch 203 and the storage device 205). The physical NIC 312 includes virtual NICs 327-1 to 327-n, a routing unit 328, and a processing unit 329.

  The virtual NICs 327-1 to 327-n have virtual NIC network functions corresponding to the virtual machines 1 (301-1) to n (301-n), respectively. That is, the virtual NICs 327-1 to 327-n have a function for executing network emulation processing in each virtual machine 301. When the physical NIC 312 executes network emulation processing, the processing performance of the server system 204 can be improved.

  The virtual NICs 327-1 to 327-n include transmission buffers 324-1 to 324-n and reception buffers 325-1 to 325-n, respectively. The transmission buffers 324-1 to 324-n are buffers that temporarily store data transmitted from each virtual machine 301. The reception buffers 325-1 to 325-n are buffers that temporarily store data received by the virtual machines 301.

  The virtual machine monitor 302 manages the virtual driver 323 in the virtual machine 301 and the virtual NIC 327 of the physical NIC 312 in association with each other. In the example illustrated in FIG. 3, the virtual driver 323-1 and the virtual NIC 327-1 are associated with each other, the virtual driver 323-2 and the virtual NIC 327-2 are associated with each other, and the virtual driver 323-n and the virtual NIC 327-n are associated with each other. Are matched.

  Hereinafter, when the virtual NICs 327-1 to 327-n are not distinguished, they are described as virtual NICs 327. Further, when the transmission buffers 324-1 to 324 -n are not distinguished from each other, they are described as transmission buffers 324, and when the reception buffers 325-1 to 325 -n are not distinguished from each other, they are described as reception buffers 325.

  The routing unit 328 outputs the transmission data transmitted from each virtual computer 301 to the processing unit 329, and performs processing for distributing the data input from the processing unit 329 to each virtual NIC 327.

  The processing unit 329 receives data from the network and transmits data to the network.

  The server system 204 includes an HBA (not shown) that is an interface used for connection between the server system 204 and the storage apparatus 205. The virtual computer 301 may include a virtual HBA (not shown).

  In this embodiment, since the physical NIC executes network emulation, the virtual NIC 327 and the routing unit 328 are provided. However, the virtual machine monitor 302 may include the virtual NIC 327. In this case, the virtual machine monitor 302 executes network emulation processing.

  Note that the functions provided in each unit may be combined into one unit.

  FIG. 5 is an explanatory diagram illustrating an example of a format of the AP information management table 308 according to the first embodiment of this invention.

  The AP information management table 308 includes an AP information address management table 800. The AP information address management table 800 manages addresses of areas in which AP operation information of the AP 321 operating on each virtual machine 301 is stored.

  Here, the area indicates a part of the address space on the memory 331, and the address stored in the AP information address management table 800 stores the head address of the area. The virtual machine monitor 302 secures an area for storing AP operation information of the AP 321 operating on each virtual machine 301 on the memory 331, and stores information described below in the area.

  The area indicated by the address of the AP information address management table 800 includes an AP name 801, AP-ID 802, NW setting information, installation date / time 808, transmission log address 809, reception log address 810, AP reception processing interval 811, and pseudo congestion. A processing log address 812 and the like are stored.

  The AP name 801 is the name of the AP 321 executed on the virtual machine 301. The AP-ID 802 is an identifier that identifies the AP 321 executed on the virtual computer 301.

  The NW setting information is network setting information in the virtual NIC 327 used by the AP 321 executed on the virtual computer 301. The NW setting information includes, for example, a MAC address 803, an IP address 804, a port 805, a reception buffer size 806, a VLAN-ID 807, and the like.

  The MAC address 803 is a MAC address assigned to the virtual NIC 327 used by the AP 321. The IP address 804 is an IP address assigned to the virtual NIC 327. The port 805 is a port number of the port in the virtual NIC 327. The reception buffer size 806 is the size of the reception buffer 325 allocated to the virtual NIC 327. The VLAN-ID 807 is an identifier of a VLAN (Virtual Local Area Network) set for the virtual NIC 327.

  The installation date and time 808 is time information such as the date and time when the AP 321 is installed in the virtual computer 301.

  The transmission log address 809 is the head address of an area in which logging information of data transmitted by the AP 321 executed on the virtual computer 301 is stored.

  The reception log address 810 is the head address of an area in which logging information of data received by the AP 321 executed on the virtual computer 301 is stored. In the area indicated by the reception log address 810, a communication partner address 820, a sequence number 821, a NIC reception time stamp 822, an AP reception processing time stamp 823, reception data 824, and the like are stored.

  The communication partner address 820 is an address of a partner with which the AP executed on the virtual computer 301 communicates. The sequence number 821 is the sequence number of the received data packet that is counted for each AP 321 by the virtual machine monitor 302. The NIC reception time stamp 822 is time information when the physical NIC 312 receives the reception data packet. The AP reception processing time stamp 823 is time information when the AP 321 receives the reception data packet. Received data 824 is a received data packet itself.

  The AP reception processing interval 811 is an interval at which the AP 321 executed on the virtual computer 301 receives data.

  The pseudo congestion processing log address 812 is an address of an area in which logging information of processing executed by the pseudo congestion processing unit 306 is stored.

  In the area indicated by the pseudo congestion processing log address 812, a determination date / time 830, a CPU 831, a memory 832, a physical NIC band 833, a virtual NIC band 834, a pseudo congestion occurrence interval 835, and the like are stored.

  The determination date and time 830 is time information such as the date and time when it is determined that pseudo congestion is necessary. The CPU 831 is information on the CPU assigned to the virtual machine 301 when it is determined that pseudo congestion is necessary. For example, the CPU information is a CPU operating rate assigned to the virtual machine 301 used by the AP 321.

  The memory 832 is information on the memory allocated to the virtual machine 301 when it is determined that pseudo congestion is necessary. For example, the memory information is the memory size of the memory allocated to the virtual machine 301 used by the AP 321.

  The physical NIC band 833 is a band assigned to the physical NIC 312. The virtual NIC band 834 is a band assigned to the virtual NIC 327. The pseudo congestion occurrence interval 835 is an interval at which pseudo congestion occurs.

  FIG. 6 is an explanatory diagram illustrating an example of a format of the configuration history management table 310 according to the first embodiment of this invention.

  The configuration history management table 310 includes a configuration information address management table 900. The configuration information address management table 900 manages addresses of work areas in which hardware configuration information of the server system 204 is stored. Note that the address stored in the configuration information address management table 900 is the head address of the work area.

  The configuration history management table 310 is updated each time a hardware change occurs after the virtual machine monitor 302 starts operating on the server system 204. That is, when the hardware is changed, an address is added to the configuration information address management table 900, and the hardware configuration information is stored in an area corresponding to the address.

  The area shown in the configuration information address management table 900 stores a setting date and time 901, a CPU 902, a memory 903, a NIC 904, an HBA 905, a virtual machine total number 906, a setting information address of the virtual machine, and the like.

  The set date 901 is time information such as date and time when the hardware is changed. The CPU 902 is information of the CPU 330 in the physical computer 300 after the hardware is changed. The memory 903 is information of the memory 331 in the physical computer 300 after the hardware is changed.

  The NIC 904 is information of the physical NIC 312 after the hardware is changed. For example, the bandwidth of the physical NIC 312 is stored. The HBA 905 is information of an HBA (Host Bus Adapter) in the physical computer 300 after the hardware is changed.

  The total number of virtual machines 906 is the number of virtual machines 301 generated on the physical machine 300.

  The setting information address of the virtual machine is an address of an area in which setting information of each virtual machine 301 generated on the physical machine 300 is stored.

  In the example shown in FIG. 6, the setting information address 907 of the virtual machine 1 and the setting information address 908 of the virtual machine 2 are stored. The setting information address 907 of the virtual machine 1 is an address of an area in which the setting information of the virtual machine 1 (301-1) is stored. The setting information address 908 of the virtual machine 2 is an address of an area where setting information of the virtual machine 2 (301-2) is stored.

  Hereinafter, the information stored in the area indicated by the setting information address 907 of the virtual machine 1 will be described as an example.

  The area indicated by the setting information address 907 of the virtual machine 1 includes a setting date and time 910, an OS 911, a CPU 912, a memory 913, a virtual NIC band 914, a virtual HBA band 915, and the like.

  The set date and time 910 is time information such as the date and time when the virtual machine 1 (301-1) was generated. The OS 911 is information on the guest OS 322-1 of the virtual machine 1 (301-1). The CPU 912 is information on the CPU assigned to the virtual machine 1 (301-1). The memory 913 is information on the memory allocated to the virtual machine 1 (301-1). The virtual NIC band 914 is a band assigned to the virtual NIC 327-1 of the virtual machine 1 (301-1). The virtual HBA band 915 is a band allocated to the virtual HBA of the virtual machine 1 (301-1).

  FIG. 7 is a sequence diagram illustrating the flow of processing executed by the server system 204 in the network system according to the first embodiment of this invention.

  The AP 321-1 is installed on the virtual machine 301-1, and the AP 321-1 is activated (400).

  When the virtual machine monitor 302 detects the activation of the AP 321-1, the virtual machine monitor 302 executes the following processing.

  The virtual machine monitor 302 acquires hardware configuration information when the AP 321-1 is installed from the configuration history management table 310, and starts logging (see FIG. 9) for the AP 321-1 (401). Specifically, the virtual machine monitor 302 measures packets transmitted and received from the AP 321-1 and stores the measurement results in the AP information management table 308. The transmission packet is measured by the transmission monitor 303, and the reception packet is measured by the reception monitor 304.

  The AP 321-1 transmits / receives a packet (402, 403). The packet transmitted by the AP 321-1 is transmitted from the virtual NIC 327-1 to the network switch 203, and the packet received from the network switch 203 is received by the AP 321-1 via the virtual NIC 327-1.

  The virtual machine monitor 302 acquires the size of the reception buffer 325-1 of the virtual NIC 327-1 used by the virtual machine 301-1, the AP reception processing interval, and the like based on the measurement result of the transmitted and received packets (404). ). The acquired information is stored in an area corresponding to the virtual machine 301-1 in the AP information management table 308.

  Here, when the physical NIC 312 is changed (405), the virtual machine monitor 302 detects the change of the physical NIC 312, acquires the information of the physical NIC 312 after the change, and calculates the occurrence interval of the pseudo congestion (406). .

  Hereinafter, a method for calculating the pseudo-congestion occurrence interval will be described by taking as an example the case where the physical NIC 312 is changed to a high performance. The physical NIC when the AP 321-1 is installed is referred to as an old physical NIC 312 and the physical NIC 312 with high performance is referred to as a new physical NIC 312.

  The virtual machine monitor 302 that has detected the change of the physical NIC 312 acquires the information of the new physical NIC 312, and the configuration history management table uses the information of the new physical NIC acquired together with other hardware information as hardware configuration information of different generations. Stored in 310.

  Next, the virtual machine monitor 302 acquires hardware configuration information before the change from the configuration history management table 310 and calculates a difference in performance between the old physical NIC 312 and the new physical NIC 312.

  For example, when the old physical NIC 312 having a speed of 100 Mbps is changed to the new physical NIC 312 having a speed of 1 Gbps, the difference in performance is calculated as 1 Gbps / 100 Mbps = 10. That is, the difference in performance between the old physical NIC 312 and the new physical NIC 312 is calculated as 10 times.

  In this case, after the physical NIC 312 is changed, the AP 321-1 may receive a data amount 10 times that before the change, which may cause an overflow.

  Therefore, the virtual machine monitor 302 acquires the AP reception processing interval 811 and the reception buffer size 806 with reference to the AP information management table 308, and generates an artificial congestion state before the reception buffer overflows. (Pseudo congestion occurrence interval 414) is calculated. The pseudo congestion occurrence interval 414 is a time that is a timing for executing the pseudo congestion processing.

  A method for calculating the pseudo congestion occurrence interval will be described later with reference to FIGS. 8A and 8B.

  The virtual machine monitor 302 starts pseudo congestion generation processing (407). Specifically, the virtual machine monitor 302 starts measuring the transmitted and received packets (408, 409) based on the calculated pseudo congestion occurrence interval 414. Further, the virtual machine monitor 302 starts counting the pseudo congestion occurrence interval 414.

  The virtual machine monitor 302 notifies the virtual NIC 327-1 of the occurrence of the pseudo congestion after the time of the pseudo congestion occurrence interval 414 minutes has elapsed (410). Specifically, it notifies the occurrence of a PAUSE command requesting the stop of packet transmission.

  The physical NIC 312 such as Ethernet has a function of requesting the stop of data transmission when a congestion state is detected. For example, the network switch 203 transmits a PAUSE command to a dedicated multicast address to interrupt data transmission to the data receiver (in this case, AP 321-1), or notifies the data transmission source of collision detection. A method of interrupting data transmission by doing so can be considered.

  In the example illustrated in FIG. 7, the virtual NIC 327-1 that has received the notification of the occurrence of pseudo congestion transmits a PAUSE command to the network switch 203 (411). Thus, the SW 203 that has received the PAUSE command transmits the PAUSE command to the transmission source of the packet, whereby the flow control of the received data can be performed.

  The network switch 203 receives the PAUSE command transmitted from the virtual NIC 327-1, transmits the received PAUSE command to the address of the data transmission source, and temporarily stops data transmission from the data transmission source (412). As a result, the amount of packets received by the AP 321-1 is kept constant. That is, the occurrence of overflow can be avoided.

  By generating pseudo congestion, the AP 321-1 can receive a packet having the same data amount as before the change to the new physical NIC 312 (413).

  In addition, since pseudo-congestion is notified to the virtual NIC 327-1 used by the AP 321-1, the AP 321 that uses another virtual NIC 327 is not affected.

  8A and 8B are flowcharts illustrating processing executed by the virtual machine monitor 302 according to the first embodiment of this invention.

  When starting the process, the virtual machine monitor 302 determines whether or not to newly create the virtual machine 301 (step 500). For example, the determination is made based on whether a generation request for the virtual machine 301 is received from the administrator of the server system 204.

  When it is determined that a new virtual machine 301 is to be generated, the virtual machine monitor 302 determines hardware configuration information such as a CPU and a memory to be allocated to the virtual machine 301, and configuration history management of the determined hardware configuration information is performed. It stores in the table 310 (step 501).

  Next, the virtual machine monitor 302 determines whether or not the AP 321 is installed in the generated virtual machine 301 (step 502). For example, the determination is made by detecting whether a user who uses the generated virtual machine 301 has installed the AP 321 in the virtual machine.

  When it is determined that the AP 321 is installed in the generated virtual machine 301, the virtual machine monitor 302 displays setting information necessary for executing the installed AP 321 (for example, the network such as the name of the AP 321 and the IP address). Setting information, etc.) is acquired and stored in the AP information management table 308 (step 503).

  Specifically, the virtual machine monitor 302 secures an area for storing information of the AP 321 executed on the virtual machine 301 generated on the memory 331, and further secures a transmission log area and a reception log area. The acquired information is stored in each secured area.

  The virtual machine monitor 302 determines whether or not the installed AP 321 is activated (step 504).

  If it is determined that the installed AP 321 is not activated, the virtual machine monitor 302 returns to step 502 and executes the same processing.

  When it is determined that the installed AP 321 is activated, the virtual machine monitor 302 starts logging for acquiring AP operation information (step 505). Details of the logging will be described later with reference to FIG.

  The virtual machine monitor 302 determines whether or not pseudo congestion occurs (step 506).

  Specifically, the virtual machine monitor 302 determines whether or not the time interval indicated by the pseudo congestion occurrence interval 835 has been reached after the pseudo congestion occurrence processing is started. Further, when the time interval indicated by the pseudo-congestion occurrence interval 835 has elapsed, the virtual machine monitor 302 detects the number of data packets buffered in the reception buffer 325 and sets the reception log address in the AP information management table 308. By referring to the information of the indicated area, it is determined whether or not the number of buffered data packets does not cause buffer over until the time interval indicated by the next pseudo congestion occurrence interval 835 elapses.

  If it is determined that the number of buffered data packets will not cause buffer over until the time interval indicated by the next pseudo-congestion occurrence interval 835 elapses, the virtual machine monitor 302 displays the next pseudo-congestion. The pseudo congestion process is not executed until the time interval indicated by the generation interval 835 has elapsed.

  When it is determined that the pseudo congestion is generated, the virtual machine monitor 302 updates the information of the virtual machine 301 in the AP information management table 308 (step 507), and notifies the virtual NIC 327 of the occurrence of the pseudo congestion (step 508). .

  Specifically, the pseudo congestion processing unit 306 acquires the current time, and stores the acquired current time in the pseudo congestion occurrence logging area of the AP information management table 308.

  In addition, the virtual NIC 327 that has received the PAUSE command transmission request transmits the PAUSE command to the network switch 203 connected to the physical NIC 312 in which the virtual NIC 327 is set. Since the PAUSE command in Ethernet includes a time for stopping transmission, it is possible to automatically restart data transmission by setting the transmission stop time.

  The virtual machine monitor 302 determines whether the AP 321 has ended (step 509). The AP 321 is terminated by the user or the administrator of the server system 204.

  If it is determined that the AP 321 has not ended, the virtual machine monitor 302 executes the processing from step 505 to step 508.

  If it is determined that the AP 321 has ended, the virtual machine monitor 302 ends the process.

  If it is determined in step 500 that the virtual computer 301 is not newly generated, or if it is determined that the AP 321 is not installed in the virtual computer 301 generated in step 502, the virtual computer monitor 302 displays the hardware of the server system 204. It is determined whether or not the wear has been changed (step 510).

  If it is determined that the hardware of the server system 204 has not been changed, the virtual machine monitor 302 returns to step 500 and executes the same processing.

  When it is determined that the hardware of the server system 204 has been changed, the virtual machine monitor 302 acquires the hardware configuration information of the server system 204 after the hardware has been changed and updates the configuration history management table 310. (Step 511).

  Specifically, the virtual machine monitor 302 secures an area for storing the hardware configuration information of the server system 204 after the change on the memory 331, and starts the reserved area in the configuration information address management table 900. Add an address.

  Further, the virtual machine monitor 302 stores the acquired information (for example, setting date / time, CPU, memory, NIC processing speed, protocol version, etc.) in the area for storing the hardware configuration information of the server system 204 after the change. ). As for the hardware that is not changed, the same hardware configuration information as that of the server system 204 before the change is stored in the area for storing the hardware configuration information of the server system 204 after the change.

  Next, the virtual machine monitor 302 executes a migration process associated with the hardware change (step 512).

  For example, when the hardware to be changed is the physical NIC 312, the virtual machine monitor 302 stops the AP 321 that uses the physical NIC 312 and migrates the network setting information after the change to the new physical NIC 312, or the AP 321. A method of using a live migration method for migrating the virtual computer 301 to a virtual environment constructed on another server system 204 or another physical computer 300 while operating is considered.

  The virtual machine monitor 302 determines whether or not the changed hardware is the physical NIC 312 (step 513).

  If it is determined that the changed hardware is not the physical NIC 312, the virtual machine monitor 302 returns to step 500 and executes the same processing. Note that processing based on the changed hardware may be executed.

  When it is determined that the changed hardware is the physical NIC 312, the virtual machine monitor 302 executes processing for eliminating the performance difference shown in steps 514 to 518. That is, processing for generating pseudo congestion is executed.

  First, the virtual machine monitor 302 acquires the changed performance information (for example, processing speed and protocol information) of the physical NIC 312 from the configuration history management table 310 (step 514).

  The virtual machine monitor 302 acquires the hardware configuration information of the server system 204 at the time when the AP 321 is installed from the AP information management table 308 and the configuration history management table 310 (step 515). That is, the performance information of the physical NIC 312 before the change is acquired. This processing is executed by the performance difference elimination processing unit 305.

  Specifically, the virtual machine monitor 302 acquires the hardware configuration information of the server system 204 at the time when the AP 321 is installed from the configuration history management table 310.

  Next, the virtual machine monitor 302 identifies the AP 321 that is executed on the virtual machine to which the physical NIC 312 before the change has been assigned, based on the setting date and time 901 of the acquired hardware configuration information. Further, the virtual machine monitor 302 acquires information on the AP 321 specified from the AP information management table 308. The specified AP 321 information includes the buffer size of the reception buffer 325 of the virtual NIC 327 and the reception processing interval of the AP 321.

  Hereinafter, the physical NIC 312 before the change is referred to as a first generation physical NIC 312, and the physical NIC 312 after the change is referred to as a second generation physical NIC 312.

  Based on the acquired performance information of the first generation physical NIC 312 and the performance information of the second generation physical NIC 312, the virtual machine monitor 302 determines the performance difference between the first generation physical NIC 312 and the second generation physical NIC 312. Is calculated (step 516). This processing is executed by the performance difference elimination processing unit 305.

  For example, when the speed of the first generation physical NIC 312 is 100 Mbps and the speed of the second generation physical NIC 312 is 1 Gbps, the second generation physical NIC 312 is calculated by calculating 1 Gbps / 100 Mbps. It can be seen that the data transmission / reception capability is 10 times that of the physical NIC 312.

  When it is determined that the performance of the second generation physical NIC 312 is higher than the performance of the first generation physical NIC 312, the virtual machine monitor 302 determines to generate pseudo congestion and executes the following processing.

  If the performance of the second generation physical NIC 312 is lower than the performance of the first generation physical NIC 312, the process may proceed to step 509 because it is not necessary to generate pseudo congestion.

  The virtual machine monitor 302 is based on the information of the AP 321 that used the virtual NIC 327 on the first generation physical NIC 312 and the calculated performance difference between the first generation physical NIC 312 and the second generation physical NIC 312. The pseudo congestion occurrence interval is determined (step 517). This processing is executed by the performance difference elimination processing unit 305.

  For example, the performance difference between the first generation physical NIC 312 and the second generation physical NIC 312 is 10 times, the reception processing interval of the AP 321 is 100 ms, and the buffer size of the reception buffer 325 of the virtual NIC 327 is 10 data packets. In some cases, an overflow may occur in the reception buffer 325 of the virtual NIC 327 10 ms after the data is received. Therefore, in this case, the virtual machine monitor 302 determines the pseudo congestion occurrence interval as “10 ms”, and stores the determined pseudo congestion occurrence interval in the pseudo congestion occurrence interval 835 of the AP information management table 308.

  The virtual machine monitor 302 starts pseudo congestion generation processing using the determined pseudo congestion generation interval (step 518).

  Thereafter, the virtual machine monitor 302 proceeds to step 505 and executes the processing from step 505 to step 509.

  In step 518, the virtual computer monitor 302 sets an upper limit value of the amount of data buffered in the reception buffer 325 of the virtual NIC 327, and the amount of data buffered in the reception buffer 325 of the virtual NIC 327 is within the pseudo-congestion occurrence interval. However, when the set upper limit value is exceeded, a process of notifying a PAUSE command transmission request may be performed.

  The virtual machine monitor 302 may determine an upper limit value set in the reception buffer 325 of the virtual NIC 327 and execute processing using the upper limit value. That is, the virtual machine monitor 302 may determine whether or not pseudo congestion has occurred based on the upper limit value of the buffer size of the reception buffer 325 of the virtual NIC 327 instead of the pseudo congestion occurrence interval.

  Further, when the amount of buffering to the reception buffer 325 of the virtual NIC 327 is large and the PAUSE command is transmitted every time the pseudo congestion occurs, for example, the reception buffer of the virtual NIC 327 of the AP 321 is further added to the virtual NIC 327. A process of increasing the amount of receivable data with 325 may be executed.

  When returning from step 509 to step 505, the virtual machine monitor 302 determines whether or not the hardware has been changed. When the hardware has been changed, the virtual machine monitor 302 executes the processing from step 511 to step 518. You may do it.

  FIG. 9 is a flowchart illustrating AP operation information logging processing executed by the virtual machine monitor 302 according to the first embodiment of this invention.

  The process shown in FIG. 9 is a process executed in step 505. Note that the processing described below is executed by the transmission monitor 303 and the reception monitor 304.

  The virtual machine monitor 302 initializes the measurement area after the AP 321 is activated in step 504 (step 600).

  The virtual machine monitor 302 acquires the setting information of the activated AP 321 from the AP information management table 308 (step 601). The acquired setting information includes, for example, address information of the communication partner, the buffer size of the reception buffer 325 of the virtual NIC 327 (reception buffer size 806), and the like.

  The virtual machine monitor 302 determines whether or not there is an interrupt from the physical NIC 312 (step 602).

  If it is determined that there is no interrupt from the physical NIC 312, the virtual machine monitor 302 returns to step 602 and executes the same processing until it detects an interrupt from the physical NIC 312.

  When it is determined that there is an interrupt from the physical NIC 312, that is, when an interrupt from the physical NIC 312 is detected, the virtual machine monitor 302 determines whether the interrupt is an interrupt for data reception (step 603).

  If it is determined that the interrupt from the physical NIC 312 is an interrupt for data reception, the virtual machine monitor 302 buffers the received data in the reception buffer 325 of the virtual NIC 327 and adds a time stamp to the received data (step 604). ). The given time stamp indicates the time at which the physical NIC received the reception data, and information on the time stamp is stored in the NIC reception time stamp 822.

  Next, the virtual machine monitor 302 calculates an interval at which received data is received (step 605).

  Specifically, the virtual machine monitor 302 acquires the current time, calculates the time interval between the acquired current time and the time when the previous data reception interrupt occurred, and the calculation result is stored in the AP information management table 308. To store.

  The virtual machine monitor 302 determines whether or not the AP 321 has executed reception processing (step 606). That is, it is determined whether or not the AP 321 has received the received data.

  If it is determined that the AP 321 is not executing reception processing, the virtual machine monitor 302 proceeds to step 608.

  If it is determined that the AP 321 is executing the reception process, the virtual machine monitor 302 adds a time stamp to the received data (step 607). The given time stamp indicates the time at which the AP 321 received the received data, and information on the time stamp is stored in the AP reception processing time stamp 823.

  Next, the virtual machine monitor 302 acquires the current time, calculates a time interval (AP reception processing interval) between the acquired current time and the time when the previous AP 321 executed the reception processing, and calculates The result is stored in the AP information management table 308 (step 608), and the process is terminated.

  In this embodiment, the physical NIC 312 is targeted as hardware managed for each generation, but other hardware such as the CPU 330 and the memory 331 may be used. For example, when the hardware to be changed is the CPU 330, the performance difference due to the hardware change is resolved by processing such as limiting the usage rate so as to approximate the processing capacity of 100% of the CPU before the change. it can.

  According to the first embodiment, the virtual machine monitor 302 records hardware configuration information of the server system 204 when an application is installed on the virtual machine 301. When the physical NIC is changed, the virtual machine monitor 302 may perform flow control of received data by generating pseudo congestion in consideration of the performance difference between the physical NIC 312 before the change and the physical NIC 312 after the change. it can. As a result, it is possible to prevent abnormal operation and abnormal termination of applications that occur due to network speedup.

[Second Embodiment]
A second embodiment of the present invention will be described with reference to FIGS.

  The second embodiment is different from the first embodiment in that the network switch 203 executes processing for generating pseudo congestion.

  That is, when the physical NIC 312 of the physical computer 300 is changed, the virtual computer monitor 302 notifies the network switch 203 connected to the physical computer 300 of a pseudo congestion occurrence interval for preventing the reception buffer from overflowing, and the network The switch 203 generates pseudo congestion according to the notification.

  Hereinafter, the difference from the first embodiment will be mainly described.

  Since the configuration of the network system of the second embodiment is the same as that of the first embodiment, description thereof is omitted. In addition, the configuration of the server system 204 and the physical computer 300 is the same as that of the first embodiment, and a description thereof will be omitted.

  FIG. 10 is a block diagram illustrating an example of the internal configuration of the network switch 203 according to the second embodiment of this invention.

  The network switch 203 includes a CPU 1100, an address memory 1101, a packet memory 1102, a switch chip 1103, a LAN (LOCAL AREA NETWORK) port 1104, and a memory 1120.

  CPU 1100 executes a program stored in memory 1120.

  The address memory 1101 stores information regarding the network address. Specifically, a MAC (Media Access Control) address table (not shown), a routing table (not shown), and the like are stored.

  The packet memory 1102 is a memory that temporarily stores data packets received by the network switch.

  The switch chip 1103 executes various network processes. Specifically, the switch chip 1103 temporarily stores the data packet received at the LAN port 1104 in the packet memory 1102, searches the address memory 1101 for the destination address LAN port 1104, and the LAN port 1104. Data packet delivery processing to the server is executed. When received data addressed to the network switch 203 itself is received, it is transferred to the CPU 1100 and processing according to the received data is executed.

  The memory 1120 stores a program executed by the CPU 1100 and information necessary for executing the program.

  The memory 1120 includes a route information transmission processing unit 1105, a route information reception processing unit 1106, a routing table management unit 1107, an AP operation guarantee processing unit 1108, an AP-network environment information table 1109, a pseudo congestion information transmission processing unit 1110, and pseudo congestion information. A reception processing unit 1111 is included.

  The route information transmission processing unit 1105 executes a transmission process of the processing result executed by the routing table management unit 1107 and data that needs to be notified to the other network switch 203.

  The route information reception processing unit 1106 outputs the received data packet to the routing table management unit 1107 when it is addressed to itself and is data related to the route information.

  The routing table management unit 1107 executes the transfer process of the received data packet according to the route information.

  When the data packet received from the server system 204 is data including AP operation information, the AP operation guarantee processing unit 1108 stores the AP operation information included in the data packet in the AP-network environment information table 1109 and receives it. Pseudo congestion processing such as monitoring of data packet reception intervals is executed.

  The pseudo congestion information reception processing unit 1111 receives information related to pseudo congestion from the server system 204. Furthermore, information regarding pseudo congestion is received from another network switch 203 or the like.

  The pseudo congestion information transmission processing unit 1110 transmits information related to pseudo congestion received from the server system 204 to another network switch 203 or the like.

  The network switch 203 may include a one-chip microcomputer including a CPU, a ROM, a RAM, and the like instead of the CPU 1100 and the memory 1120.

  FIG. 11 is a sequence diagram illustrating the flow of processing executed by the server system 204 and the network switch 203 in the network system according to the second embodiment of this invention.

  The processes after starting the AP 321-1 installed in the virtual machine 301-1 (step 400, step 401) and the process for changing the physical NIC 312 (step 405) are the same as those in FIG.

  The virtual machine monitor 302 that has detected that the physical NIC 312 has been changed acquires information on the physical NIC 312 at the time when the AP 321-1 was installed from the configuration history management table 310 (1200). Specifically, the bandwidth information of the physical NIC 312 at the time when the AP 321-1 is installed is acquired from the NIC 904 of the configuration history management table 310.

  The virtual machine monitor 302 transmits the acquired bandwidth information of the physical NIC 312 to the virtual NIC 327-1 (1201). Further, the bandwidth information of the physical NIC 312 obtained is transmitted to the network switch 203 via the virtual NIC 327-1 (1202).

  The virtual machine monitor 302 acquires the reception processing interval of the AP 321-1 from the AP information management table 308, and calculates an AP receivable bandwidth that is a receivable bandwidth of the AP 321-1 (1203). The AP receivable bandwidth is calculated, for example, by multiplying the reception processing interval of the AP 321 and the reception data size.

  AP operation information including the calculated AP receivable bandwidth, the acquired reception processing interval of the AP 321-1, the network address assigned to the virtual NIC 327-1, and the like is transmitted to the virtual NIC 327-1 (1204). Further, AP operation information is transmitted to the network switch 203 via the virtual NIC 327-1 (1205).

  The bandwidth information and AP operation information of the physical NIC transmitted from the virtual machine monitor 302 to the network switch 203 may be included in the same data packet and transmitted at a time.

  The network switch 203 that has received the AP operation information acquires a reception processing interval, an AP receivable band, and a network address included in the received AP operation information (1206).

  The network switch 203 measures the reception interval of the data packet for the acquired network address (data transmission source 1220) (1207).

  The network switch 203 determines whether or not pseudo congestion processing is necessary based on the measurement result (1208). Note that a method using the same method as 406 in FIG.

  If it is determined that pseudo congestion processing is necessary, the network switch 203 transmits a PAUSE command to the data transmission source 1220 in order to temporarily stop transmission of the data packet for the acquired network address (1209).

  FIG. 12 is a flowchart illustrating processing for eliminating the performance difference executed by the virtual machine monitor 302 according to the second embodiment of this invention.

  Of the processes executed by the virtual machine monitor 302, the processes from step 500 to step 513 are the same, and the description thereof is omitted.

  The virtual machine monitor 302 acquires the changed performance information (for example, processing speed and protocol information) of the physical NIC 312 from the configuration history management table 310 (step 514).

  The virtual machine monitor 302 acquires the hardware configuration information of the server system 204 at the time when the AP 321 is installed from the AP information management table 308 and the configuration history management table 310 (step 515). Specifically, the bandwidth information of the physical NIC at the time when the AP 321-1 is installed is acquired.

  The virtual machine monitor 302 transmits the acquired bandwidth information of the physical NIC 312 to the network switch 203 (step 1300). Note that the bandwidth of the virtual NIC 327 may be transmitted as information of the physical NIC 312 in the server system 204 at the time when the AP 321 is installed.

  The virtual machine monitor 302 acquires the AP 321-1 reception processing interval, reception buffer size, and the like from the AP information management table 308 (step 1301).

  The virtual machine monitor 302 calculates the AP receivable bandwidth based on the acquired reception processing interval of the AP 321-1 (step 1302).

  The virtual machine monitor 302 transmits AP operation information including the calculated AP receivable bandwidth, the acquired reception processing interval of the AP 321-1 and the network address assigned to the virtual NIC 327-1 to the network switch 203. (Step 1303).

  Thereafter, the AP operation information logging process of step 505 is executed until the AP 321-1 is terminated without executing the processes of step 506 to step 508.

  FIG. 13 is a flowchart illustrating processing executed by the network switch 203 according to the second embodiment of this invention.

  The network switch 203 determines whether a data packet is received via the LAN port 1104 (step 1400).

  If it is determined that the data packet has not been received, the network switch 203 returns to step 1400 and continues to wait until the data packet is received.

  When it is determined that the data packet has been received, the network switch 203 stores the received data packet in the packet memory 1102, and determines whether the received data packet is addressed to itself (network switch 203) ( Step 1401).

  If it is determined that the received data packet is not addressed to itself, the network switch 203 executes a switching process (step 1404) and returns to step 1400. The details of the switching process will be described later with reference to FIG.

  When it is determined that the received data packet is addressed to itself, the network switch 203 determines whether or not the received data packet is data related to route information (step 1402).

  If it is determined that the received data packet is data related to the route information, the network switch 203 executes routing information processing (step 1403). This processing is executed by the routing table management unit 1107.

  When it is determined that the received data packet is not data related to route information, the network switch 203 determines whether the received data packet is AP operation information (step 1405).

  If it is determined that the received data packet is not AP operation information, the network switch 203 returns to Step 1400.

  When it is determined that the received data packet is AP operation information, the network switch 203 determines from the received data packet the AP receivable bandwidth, the bandwidth information of the physical NIC 312 before the change in the AP 321-1, and the reception of the AP 321-1. A processing interval or the like is acquired (step 1406).

  Based on the information acquired from the AP operation information, the network switch 203 starts a data packet monitoring process for the virtual machine 301-1 that is the data packet transmission source (step 1407).

  FIG. 14 is a flowchart illustrating the switching process executed by the network switch 203 according to the second embodiment of this invention.

  The switch chip 1103 of the network switch 203 acquires communication destination address information included in the received data packet, and searches a routing table (not shown) stored in the address memory 1101 (step 1500).

  The network switch 203 executes a process for determining whether or not the received data packet is a data packet transmitted to the monitoring target virtual machine 301-1 based on the search result (step 1501).

  The network switch 203 determines whether or not the data packet received as a result of the determination process is a data packet transmitted to the virtual machine 301-1 to be monitored (step 1502).

  If it is determined that the received data packet is not a data packet transmitted to the monitored virtual machine 301-1, the network switch 203 returns to step 1502.

  When it is determined that the received data packet is a data packet transmitted to the virtual machine 301-1 to be monitored, the network switch 203 acquires the current time and assigns a time stamp to the received data packet. Also, the number of received data packets is counted up (step 1503).

  The network switch 203 calculates a reception interval, which is an interval at which data packets are received, and confirms the number of data packets addressed to the virtual machine 301-1 of the received data packets (step 1504).

  Specifically, the network switch 203 calculates the reception interval based on the time when the previous data packet was received and the time acquired in step 1503. The number of data packets is counted down when the data packet is output to the transmission destination, and the number of data packets buffered in the packet memory 1102 is managed for each transmission destination.

  Based on the calculated data packet reception interval and the number of data packets, the network switch 203 determines whether or not the AP 321-1 is within the reception processable range (step 1505).

  For example, when the bandwidth of the physical NIC 312 (or virtual NIC 327-1) before the change is 100 Mbps, the reception processing interval is 100 ms, and the bandwidth of the physical NIC 312 after the change is 1 Gbps, the data packet received by the network switch 203 The reception interval is calculated as 10 ms. At this time, if the number of data packets is 10, the overflow may occur in the reception buffer 325-1 of the virtual NIC 327-1 of the virtual machine 301-1. It is determined. That is, it is determined that pseudo congestion needs to be generated.

  When it is determined that the AP 321-1 is within the reception processable range, the network switch 203 ends the process.

  When it is determined that the AP 321-1 is not within the reception processable range, the network switch 203 notifies the communication partner that transmits the data packet to the AP 321-1 of the occurrence of pseudo congestion (Step 1506), and ends the process. To do.

  Specifically, the network switch 203 transmits a PAUSE command to a communication partner that transmits a data packet to the AP 321-1.

  Thereby, the transmission of the data packet from the communication partner that transmits the data packet to the AP 321-1 can be temporarily stopped.

  FIG. 15 is an explanatory diagram showing an example of the format of the AP-network environment information table 1109 managed by the network switch 203.

  The AP-network environment information table 1109 includes a port 1601, a MAC address 1602, an IP address 1603, a speed 1604, an AP reception interval 1605, an AP reception data number 1606, a reception time 1607, and a reception interval 1608.

  A port 1601 is a port number of the LAN port 1104. The MAC address 1602 is a MAC address assigned to the virtual machine 301 to be monitored. The MAC address is assigned by the virtual machine monitor 302.

  An IP address 1603 is an IP address assigned to the virtual machine 301 to be monitored. The IP address is assigned by the virtual machine monitor 302.

  The speed 1604 is a band of the physical NIC 312 (or virtual NIC 327) used by the AP 321. The AP reception interval 1605 is an interval at which the AP 321 can receive data. The AP received data number 1606 is the number of data packets addressed to each virtual machine 301 buffered in the network switch 203.

  A reception time 1607 is a time when the network switch 203 receives a data packet addressed to each virtual machine 301. A reception interval 1608 is a reception interval of data packets addressed to each virtual machine 301.

  In the second embodiment, the AP receivable bandwidth is calculated by the virtual machine monitor 302. However, the network switch 203 sets the AP receivable bandwidth based on the reception processing interval of the AP 321-1 included in the AP operation information. It may be calculated. As a result, the processing load on the virtual machine monitor 302 can be reduced.

  According to the second embodiment, when the physical NIC 312 is changed, the network switch 203 performs the performance of the physical NIC 312 before the change and the physical NIC 312 after the change based on information transmitted from the virtual machine monitor 302. The flow control of received data can be performed by generating pseudo-congestion in consideration of the difference. As a result, it is possible to prevent abnormal operation and abnormal termination of applications that occur due to network speedup.

100 Software 101 Virtual machine monitor 102 Virtual machine 106 Application operation information 107, 108 Hardware 109 Hardware performance difference information 110 Performance difference elimination processing unit 111 Pseudo congestion information reception processing unit 200 DC
201 External network 202 Router 203 Network switch 204 Server system 205 Storage device 206 Network 300 Physical computer 301 Virtual computer 302 Virtual computer monitor 303 Transmission monitor 304 Reception monitor 305 Performance difference elimination processing unit 306 Pseudo congestion processing unit 307 AP information management unit 308 AP Information management table 309 Configuration history management unit 310 Configuration history management table 312 Physical NIC
320 Internal bus 321 AP
322 Guest OS
323 Virtual driver 324 Transmission buffer 325 Reception buffer 327 Virtual NIC
328 Routing unit 329 Processing unit 330 CPU
331 Memory 332 Nonvolatile storage medium 1100 CPU
1101 Address memory 1102 Packet memory 1103 Switch chip 1104 LAN port 1105 Routing information transmission processing unit 1106 Routing information reception processing unit 1107 Routing table management unit 1108 AP operation guarantee processing unit 1109 AP-network environment information table 1110 Pseudo congestion information transmission processing Unit 1111 pseudo-congestion information reception processing unit 1120 memory

Claims (15)

A virtual machine monitor that includes a processor and a memory connected to the processor, and that is executed by the processor and manages a plurality of virtual machines generated by virtually dividing a physical resource of a server; and each of the virtual machines A server that provides a virtual environment with an operating system running on and running an application,
The virtual machine monitor is
A physical resource information management unit for managing information on the physical resource allocated for each virtual calculation period;
For each virtual machine, an application information management unit that manages information related to the operating state of the application executed on the operating system;
When the performance information of the physical resource of the server is managed and the first physical resource of the server is changed to a second physical resource with high performance, the performance of the first physical resource and the second physical resource Obtaining performance difference information representing a difference from the performance of the resource, and generating control information for adjusting the performance of the second physical resource so that the application operates normally based on the performance difference information And a performance difference information management unit.   The server according to claim 1, wherein the physical resource is a network interface for the server to communicate with another device. The virtual machine monitor virtually divides the network interface and assigns a virtual network interface to each virtual machine,
The virtual network interface includes a reception buffer that temporarily accumulates data received by the application, and a transmission buffer that temporarily accumulates data transmitted by the application,
The performance difference information management unit
Holding performance information of a first network interface provided in the server;
When the first network interface is changed to a second network interface with high performance, the performance information of the second network interface is acquired;
Based on the performance information of the first network interface held by the performance difference information management unit and the acquired performance information of the second network interface, the performance of the first network interface and the second Calculate the difference with the performance of the network interface,
Generating control information for preventing the reception buffer from exceeding a capacity that can be stored based on the difference between the calculated performance of the first network interface and the performance of the second network interface; The server according to claim 2. The control information generated by the performance difference information management unit is congestion notification setting information for notifying that a congestion state has occurred in a virtual network interface allocated to a virtual machine on which the application is executed,
The server according to claim 3, wherein the virtual machine monitor further includes a congestion processing unit that notifies that a congestion state has occurred in the virtual network interface based on the congestion notification setting information. The server is connected to a network including a plurality of network switches;
The virtual machine monitor measures data reception status in the application,
The application information management unit manages the buffer size of the reception buffer acquired based on the data reception status measurement result in the application, and the application reception interval at which the application receives data,
The performance difference information management unit
Based on the performance difference information, the buffer size of the reception buffer, and the application reception interval, generate the congestion notification setting information,
The congestion processing unit
Periodically notifies the virtual network interface that the congestion state has occurred,
The server according to claim 4, wherein the virtual network interface transmits a command for stopping transmission of data addressed to the application to the network switch. The server is connected to a network including a plurality of network switches;
The virtual machine monitor measures data reception status in the application,
The application information management unit manages the buffer size of the reception buffer acquired based on the data reception status measurement result in the application, and the application reception interval at which the application receives data,
The performance difference information management unit
Based on the performance difference information, the buffer size of the reception buffer, and the application reception interval, generate threshold information indicating an upper limit value of the buffer size of the reception buffer,
The congestion processing unit
Determining whether the buffer size of the reception buffer is equal to or greater than the upper limit value indicated in the generated threshold information;
When the buffer size of the reception buffer is equal to or larger than the upper limit value indicated in the generated threshold information, the virtual network interface is notified that the congestion state has occurred,
The server according to claim 4, wherein the virtual network interface transmits a command for stopping transmission of data addressed to the application to the network switch. The control information generated by the performance difference information management unit is congestion notification setting information for notifying that a congestion state has occurred in a virtual network interface allocated to a virtual machine on which the application is executed,
The server is connected to a network including a plurality of network switches;
The virtual machine monitor measures data reception status in the application,
The application information management unit manages the buffer size of the reception buffer acquired based on the data reception status measurement result in the application, and the application reception interval at which the application receives data,
The performance difference information management unit
Based on the performance difference information, the buffer size of the reception buffer, and the application reception interval, generate the congestion notification setting information,
The server according to claim 3, wherein transmission of data addressed to the application is stopped by transmitting the generated congestion notification setting information to the network switch. The performance information of the first network interface and the performance information of the second network interface are a processing speed of the first network interface and a processing speed of the second network interface,
The performance difference information management unit
The server according to claim 3, wherein the performance difference information is calculated by dividing a processing speed of the second network interface by a processing speed of the first network interface. A computer system in which a plurality of computers are connected via a network including a plurality of network switches,
The computer includes a processor and a memory connected to the processor, and is executed by the processor, and a virtual machine monitor that manages a plurality of virtual computers generated by virtually dividing physical resources of the computer An operating system that executes on each virtual machine and executes an application,
The virtual machine monitor is
Managing physical resource information that is information related to the physical resource allocated for each virtual computation period;
For each virtual machine, manage application operation information that is information related to the operation state of the application executed on the operating system,
When the first physical resource of the computer is changed to a high-performance second physical resource, performance difference information indicating a difference between the performance of the first physical resource and the performance of the second physical resource is displayed. Acquired,
Transmitting the physical resource information, the application operation information, and the performance difference information to the network switch;
The network switch is
Generating control information for adjusting the performance of the second physical resource so that the application operates normally based on the received physical resource information, the application operation information, and the performance difference information; A featured computer system.   The computer system according to claim 9, wherein the physical resource is a network interface for the computer to communicate with another device. The virtual machine monitor virtually divides the network interface and assigns a virtual network interface to each virtual machine,
The virtual network interface includes a reception buffer that temporarily accumulates data received by the application, and a transmission buffer that temporarily accumulates data transmitted by the application,
The virtual machine monitor is
When the first network interface is changed to a high-performance second network interface, the performance information of the second network interface is acquired;
Based on the performance information of the first network interface held by the performance difference information management unit and the acquired performance information of the second network interface, the performance of the first network interface and the second network Calculate the difference from the interface performance,
Transmitting the difference between the calculated performance of the first network interface and the performance of the second network interface to the network switch;
The network switch is
Generating control information for preventing the reception buffer from exceeding a capacity that can be stored based on the difference between the calculated performance of the first network interface and the performance of the second network interface; The computer system according to claim 10. The control information is congestion notification setting information for notifying a command for the network switch to stop transmission of data addressed to the application,
The computer system according to claim 11, wherein the network switch notifies that a congestion state has occurred in the virtual network interface based on the congestion notification setting information. The virtual machine monitor manages the buffer size of the reception buffer and the application reception interval at which the application receives data as information on the operating state of the application,
The network switch is
Based on the performance difference information, the buffer size of the reception buffer, and the application reception interval, generate the congestion notification setting information,
Periodically, determine whether to generate the congestion state,
The computer system according to claim 11, wherein when it is determined that the congestion state is generated, an instruction to stop transmission of data addressed to the application is transmitted to a transmission source of the data. The performance information of the first network interface and the performance information of the second network interface are a processing speed of the first network interface and a processing speed of the second network interface,
The virtual machine monitor is
12. The computer system according to claim 11, wherein the performance difference information is calculated by dividing the processing speed of the second network interface by the processing speed of the first network interface. A virtual machine monitor that includes a processor and a memory connected to the processor, and that is executed by the processor and manages a plurality of virtual machines generated by virtually dividing physical resources of the server; A virtual machine management method in a server comprising an operating system that is executed on a computer and executes an application,
The virtual machine monitor manages information related to the physical resource allocated for each virtual calculation period, and manages information related to the operating state of the application executed on the operating system for each virtual machine. ,
The method
The virtual machine monitor is
When the first physical resource of the server is changed to a high-performance second physical resource, performance difference information indicating a difference between the performance of the first physical resource and the performance of the second physical resource is displayed. A step to obtain,
Generating virtual machine control information for reducing the performance of the second physical resource based on the acquired performance difference information so that the application operates normally, Method.

Images