WO2010064411A1

WO2010064411A1 - Device and method for communicating between virtual machines, socket api wrapper library, and storage medium

Info

Publication number: WO2010064411A1
Application number: PCT/JP2009/006516
Authority: WO
Inventors: 森本昌治
Original assignee: 日本電気株式会社
Priority date: 2008-12-02
Filing date: 2009-12-01
Publication date: 2010-06-10
Also published as: JPWO2010064411A1

Abstract

Disclosed is a system that closely links virtual machines upon a virtual machine monitor. The system facilitates rapid communication between virtual machines comprising response agents, without changing the operating systems or applications that run on the virtual machines. Also provided is a device for communicating between virtual machines, which, when a plurality of virtual machines (201 – 203) are caused to run upon a common virtual machine monitor (100), communicates between the virtual machines (201 – 203). A socket API wrapper library (socket wrapper) (501) upon the virtual machine (201) comprises a communications channel administration unit (510), which administers the communications channels provided by the virtual machine monitor (100), and a response agent unit (520), which responds to communications between the virtual machines via the communications channels.

Description

Virtual machine communication apparatus, method, socket API wrapper library, and recording medium

The present invention relates to an inter-virtual machine communication apparatus and method, and a socket API (Application Program Interface) wrapper library, and more particularly to speeding up communication between virtual machines.

Usually, the communication overhead between virtual machines (hereinafter referred to as VMs) is large, and it is known that the method using inter-VM communication significantly reduces the performance of the entire system as compared with the case where VMs are not used. In addition, when communication between VMs is frequently performed in a system in which a plurality of VMs are closely linked, if one VM is stopped due to migration or the like, the influence is propagated to other VMs, and the system May have a major impact on the operation of the system.

Therefore, a problem of a system in which a plurality of VMs closely cooperate on a virtual machine monitor (hereinafter referred to as VMM) is to achieve both high-speed communication between VMs and improvement of system availability. Further, it is important that these functions can be realized without changing the operating system (hereinafter referred to as OS) or application in the VM.

Acceleration of communication between VMs can be realized by directly using the communication channel of the VMM without going through the VM that manages communication called management VM. In particular, XenLoop (Non-Patent Document 1) and XWay (Non-Patent Document 2) have been proposed as an inter-VM communication method that can be used without changing an OS or an application using a socket.

XenLoop is a driver realized between the IP layer and the data link layer of the communication protocol stack. Therefore, XenLoop receives the data transmitted from the application in the form of a packet. If the transmission destination computer is a VM existing on the same VMM, the transmission is performed using the communication channel of the VMM. Otherwise, the packet is directly passed to the data link layer.

XWay is a switch realized between the socket API layer and the transport layer of the communication protocol stack. When data received from the socket API is transmitted to a VM in the same VMM, communication is performed using the communication channel of the VMM using the XWay protocol. Otherwise, the data is passed to the transport layer.

If these are used, high-speed communication between VMs can be realized without changing the application using the OS or socket.

On the other hand, in order to improve system availability, it is necessary to deal with a case where a VM becomes unresponsive due to migration or the like. For example, when the VM 1 transmits a request to the VM 2, the VM 1 transmission process is blocked unless the VM 2 enters the migration process and responds. Therefore, the VM1 consumes resources to save the response waiting state, or the VMM consumes resources for the communication channel.

∙ Proxy response is effective to deal with this problem. Furthermore, it is desirable to be able to respond flexibly at the application level. This is because a coping method suited to the application installed in the VM to be linked is necessary. For example, for an HTTP GET access during migration, there are applications that want to return a status code 503 indicating that the service cannot be used, and there are applications that want to return a status code 307 that is temporarily redirected to another resource.

However, in order to realize a flexible proxy response at the application level, it is necessary to modify the application itself, resulting in a problem of application development cost. Therefore, by incorporating an application level proxy response function into the mechanism of communication between VMs, it is possible to reduce the development cost of the application. In other words, an inter-VM communication method capable of proxy response that achieves both high-speed communication between VMs and improved system availability without changing the OS or application on the VM is desired.

However, if such a proxy response function is to be realized by XenLoop, XenLoop receives data in a packet format, so that it is necessary to reconstruct the packet and interpret the message. Therefore, there is a problem that the execution cost and the development cost of the proxy response function increase due to the packet restructuring process that was originally unnecessary.

On the other hand, when implemented with XWay, it is easier to interpret messages than with XenLoop, but since code is embedded in the protocol stack, there is a problem that the OS needs to be improved and the development cost increases.

Also, when considering the operation of such a system, there is a problem that work for improving the availability of the entire system by proxy response and work for improving system performance by VM migration or the like are not separated. For example, assume that a VMM administrator who manages VM placement wants to migrate VM1 placed in VMM1 of physical machine 1 onto VMM2 in physical machine 2 for performance improvement.

At that time, the VMM administrator stops the access to the VM1 from another VM (for example, VM2) in the VMM1 before starting the migration, and resumes the communication as the communication to the physical machine 2 after the migration is completed. Setting work is required. This means that both VMM administrator and service operator work are required during VM migration, leading to an increase in system operation costs. Therefore, it is important that the work of improving availability and performance can be separated.

The first problem of the above-described technology is that in a system in which VMs on a VMM are closely linked, an application using an OS or socket on the VM is changed when speeding up communication between VMs and improving system availability. Is not to.

The second problem is that a flexible proxy response at the application level is possible in order to increase the availability of the entire system in the above system.

The third problem is that the above-described system can independently perform the work of improving availability and improving performance (VM migration).

In view of the above circumstances, the present invention provides a high-speed inter-virtual machine communication having a proxy response without changing the operating system or application in the virtual machine in a system that closely links the virtual machines on the virtual machine monitor. It aims to be realized.

In order to achieve the above object, one aspect of the inter-virtual machine communication device of the present invention includes a virtual machine monitor and a plurality of virtual machines that are executed on the virtual machine monitor and communicate with each other, At least one of the plurality of virtual machines includes a wrapper library of socket API (Application Program Interface). The socket API wrapper library includes a communication channel management unit that manages a communication channel provided by the virtual machine monitor, and a proxy response unit that performs a proxy response of communication between virtual machines using the communication channel. .

An aspect of the inter-virtual machine communication method according to the present invention is an inter-virtual machine communication method for performing communication between a plurality of virtual machines on the same virtual machine monitor, and includes at least one of the plurality of virtual machines. One wrapper API wrapper library manages a communication channel provided by the virtual machine monitor, and performs a proxy response for communication between virtual machines using the communication channel.

One aspect of the socket API wrapper library according to the present invention is a communication channel management means for managing a communication channel provided by the virtual machine monitor for a virtual machine executed on the virtual machine monitor, and the communication channel. It is made to function as a proxy response means for performing a proxy response of the used inter-virtual machine communication.

One aspect of the computer-readable recording medium according to the present invention includes a communication channel management procedure for managing a communication channel provided by the virtual machine monitor for a virtual machine executed on the virtual machine monitor, and the communication A program for executing a proxy response procedure for performing a proxy response for communication between virtual machines using a channel is stored.

According to the present invention, in a system in which virtual machines on a virtual machine monitor are closely linked, high-speed communication between virtual machines having a proxy response can be realized without changing the operating system or application in the virtual machine. It becomes possible.

It is a block diagram which shows the whole structure containing the form for implementing this invention. It is a block diagram which shows the structure of the form for implementing this invention. It is a figure which shows the specific example of the VM information management table of the form for implementing this invention. It is a figure which shows the specific example of the communication channel management table of the communication channel management part of the form for implementing this invention. It is a figure which shows the specific example of the application policy management table of the proxy response part of the form for implementing this invention. It is a figure which shows the specific example of the proxy response module group of the form for implementing this invention. It is a flowchart figure which shows the operation | movement of the form for implementing this invention.

First, an outline of the present embodiment will be described.
The inter-VM communication method according to the present embodiment described below includes a communication channel management unit having a VM information management table and a communication channel management table, and a proxy response unit having an application policy management table and a proxy response module group. This is implemented as a socket API wrapper library (hereinafter referred to as “socket wrapper”).

By adopting such a configuration, it is possible to realize high-speed communication between VMs having a proxy response without changing the OS or application in the VM. Furthermore, in the system according to the present embodiment, the proxy response is automatically performed when the policy is met, so the VMM administrator can freely perform migration without notifying the service operator.
Hereinafter, the present embodiment will be described in detail with reference to the drawings.

Referring to FIG. 1, the present embodiment is realized as socket wrappers (501 to 503) linked to applications (301 to 303) by VMs (201 to 203) on the VMM (100).

Note that the present embodiment is realized by a software program that performs predetermined information processing using a general-purpose computer including an arithmetic / processing device such as a CPU (Central Processing Unit) or a processor, and a storage device such as a memory. . Such a software program and a computer-readable recording medium storing the software program also constitute the present invention.

Each application communicates with other applications through the socket API. The VMM (100) provides communication channels (701, 702) as a communication mechanism between VMs. For example, in FIG. 1, an application A (301) calls an API of a socket wrapper, and communicates with an application B (302) via a communication channel (701) provided by the VMM (100). Similarly, application B (302) communicates with application C (303) via communication channel (702).

Next, the configuration of the socket wrappers 501 to 503 will be described. Referring to FIG. 2, the socket wrapper 501 includes a communication channel management unit 510 and a proxy response unit 520. The communication channel management unit 510 has a VM information management table 511 and a communication channel management table 512. The proxy response unit 520 has an application policy management table 521 and a proxy response module group 522. Hereinafter, each table or module group shown in FIG. 2 will be described with reference to specific examples in FIG. 3 to FIG. 6, but these are merely examples for description and do not limit the present invention.

FIG. 3 shows an example of the VM information management table 511. In this table, the IP address of the destination VM and the VM identifier (hereinafter referred to as VM-ID) are managed as a set of entries. An entry of this table is added when the migration of the VM to the VMM is completed by migration, and is deleted when the migration of another VM is completed by migration. Therefore, it can be determined with reference to the VM information management table 511 whether or not an arbitrary VM (for example, a destination VM) exists on the same VMM. Here, the same VMM means one VMM of an arbitrary physical machine. Therefore, an arbitrary VM exists on the same VMM, an arbitrary VM and its own VM exist on one VMM.

FIG. 4 shows an example of the communication channel management table 512. In this table, a set of a socket descriptor number, a channel number of a VMM to be used, or a channel ID indicating a module identifier of a proxy response when proxy response is performed, and a VM-ID which is a VM identifier of a communication partner As an entry. The module identifier and channel number do not overlap.

FIG. 5 shows an example of the applied policy management table 521. In this table, the VM identifier VM-ID of the communication partner, the threshold value indicating the upper limit number of channels that is a condition for applying the proxy response, the counter indicating the number of channels being used, and the module identifier that actually performs proxy response processing ( And (Module ID) are managed as a set of entries.

FIG. 6 shows an example of the proxy response module group 522. In this table, a module identifier (Module ID) and an address where the module is arranged are managed as a set of entries. In the setting of the proxy response module, it is possible to arrange functions suitable for services and situations.

Next, the overall operation of the present embodiment will be described in detail with reference to the flowcharts of FIGS. 1, 2, 4, 5, and 7.

First, the application A301 calls a connection request API (for example, connect ()) of the socket wrapper 501 in order to connect to a communication partner (step S101 in FIG. 7).

Next, the socket wrapper 501 confirms the VM information management table 511 of the communication channel management unit 510 and determines whether the communication partner exists on the same VMM using the destination IP address as a key (an entry exists in the VM information management table 511). Search) (step S102).

If they do not exist on the same VMM, the socket wrapper 501 performs communication by calling a normal socket API (steps S103 to S104).

If they exist on the same VMM, the socket wrapper 501 acquires the VM-ID from the VM information management table 511, and searches the application policy management table 521 of the proxy response unit 520 using the VM-ID as a key (step S105). .

As a result, if the proxy response application condition (“counter> threshold” in the corresponding VM-ID entry in FIG. 5) is not satisfied, the socket wrapper 501 secures the communication channel of the VMM, and the communication channel management table. An entry (descriptor number, channel ID, VM-ID) is added to 512 and communication processing is started (steps S106 and S107).

If the proxy response application condition is satisfied, the socket wrapper 501 acquires the proxy response module (Module ID) specified in the application policy management table 521, and the entry (descriptor number, Module) in the communication channel management table 512. (ID, VM-ID) is added and proxy response processing is started (steps S106 and S108).

Next, processing at the time of VM migration in this embodiment will be described.

Now, it is assumed that VM2 starts moving to a VMM on another physical machine by migration in a system in which VM1, VM2, and VM3 are linked. Then, since connection to VM2 becomes impossible, the proxy response process is started at the stage where the proxy response application conditions are met in the socket wrappers of VM1 and VM3.

When the migration of the VM2 (202) is completed, the communication channel management unit 510 that has received the migration completion event from the VMM 100 deletes the entry in the VM information management table 511. At the same time, the entry corresponding to VM2 in the communication channel management table 512 and the applied policy management table 521 is deleted. Thereby, the connection request from VM1, VM3 to VM2 is switched to the access using the normal socket API (refer to the flow of steps S103 and S104 in FIG. 7).

On the other hand, when VM4 is added to the same VMM by migration, the communication channel management unit 510 that has received the migration completion event from the VMM adds the information of VM4 as an entry to the VM information management table 511.

Next, the effect of this embodiment will be described.

The first effect is that the application using the OS and socket on the VM is not changed when speeding up communication between VMs and improving system availability. The reason is that the present invention is implemented as a wrapper for the socket API.

The second effect is that the availability of the system is increased when a certain VM cannot respond due to migration or the like. The reason is that the present invention includes a proxy response mechanism that can flexibly cope with an application level, and prompts the application to release resources. As a result, the influence on other VMs and the entire system can be minimized.

The third effect is that work for improving system availability and performance improvement by VM migration can be performed independently. This is because the system automatically performs a proxy response at the time of migration because the present invention has a proxy response function in the inter-VM communication mechanism. Therefore, the VMM administrator can freely migrate the VM without contacting the service builder.

As mentioned above, although each embodiment of this invention was described, it is not limited to said each embodiment, A various deformation | transformation is possible in the range which does not deviate from the summary.

For example, the operation (the operation shown in the flowchart) in the above-described embodiment can be executed by hardware, software, or a combined configuration of both.

When executing processing by software, a program in which a processing sequence is recorded may be installed and executed in a memory in a computer incorporated in dedicated hardware. Or you may install and run a program in the general purpose computer which can perform various processes.

For example, the program can be recorded in advance on a hard disk or ROM (Read Only Memory) as a recording medium. Alternatively, the program is temporarily or permanently stored on a removable recording medium such as a CD-ROM (Compact Disc Only Memory), MO (Magneto Optical) disk, DVD (Digital Versatile Disc), magnetic disk, or semiconductor memory. It can be stored (recorded). Such a removable recording medium can be provided as so-called package software.

Note that the program may be installed on the computer from the removable recording medium as described above, or may be wirelessly transferred from the download site to the computer. Or you may wire-transfer to a computer via networks, such as LAN (Local Area Network) and the internet. The computer can receive the transferred program and install it on a recording medium such as a built-in hard disk.

In addition to being executed in time series in accordance with the processing operations described in the above embodiment, the processing capability of the apparatus that executes the processing, or a configuration to execute in parallel or individually as necessary Is also possible.

The present invention has been described above with reference to the embodiment, but the present invention is not limited to the above. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the invention.

This application claims priority based on Japanese Patent Application No. 2008-307248 filed on Dec. 2, 2008, the entire disclosure of which is incorporated herein.

The present invention can be applied to a system that performs communication between VMs using a socket on a VMM. More specifically, the present invention can be applied to a system that links VMs that provide Web services. In addition, if a socket is used, it can be applied to a service that communicates using an application protocol (for example, HTTP, SMTP, POP, FTP, SIP, DNS, Telnet, etc.).

100 Virtual Machine Monitor (VMM)
201 to 203 Virtual machines (VM) 1 to 3
301 to 303 Applications A to C
401 to 403 Operating System (OS) A to C
501 to 503

Socket wrapper

701, 702 Communication channel in VMM 510 Communication channel management unit 511 VM information management table 512 Communication channel management table 520 Proxy response unit 521 Applicable policy management table 522 Proxy response module group

Claims

A virtual machine monitor,
A plurality of virtual machines that are executed on the virtual machine monitor and communicate with each other;
At least one of the plurality of virtual machines includes a wrapper library of socket API (Application Program Interface),
The socket API wrapper library is
Communication channel management means for managing a communication channel provided by the virtual machine monitor;
A proxy response unit configured to perform a proxy response for communication between virtual machines using the communication channel.
The communication channel management means includes a communication channel management table for managing an entry in which a communication destination virtual machine and a communication channel or a proxy response module are associated with a socket descriptor.
The communication channel management table is referred to during communication between the plurality of virtual machines, and when the proxy response module is associated, the proxy response unit performs a proxy response. Item 2. The virtual machine communication device according to Item 1.
The communication channel management means, when migrating a virtual machine,
If the migration is to delete the virtual machine from the virtual machine monitor, delete the entry related to the virtual machine from the communication channel management table;
The virtual machine according to claim 2, wherein, when the migration is to add the virtual machine on the virtual machine monitor, an entry related to the virtual machine is added to the communication channel management table. Inter-machine communication device.
The proxy response means includes an application policy management table that manages an application policy serving as a reference for determining whether to apply a proxy response in communication between virtual machines.
When performing communication using the communication channel, the socket API wrapper library determines whether to apply a proxy response based on information in the applied policy management table. The virtual machine according to any one of claims 1 to 3, wherein when a proxy response module that executes a response executes a proxy response and determines that the response is not applied, communication is performed using the communication channel. Intercommunication device.
The applied policy management table stores the upper limit number of communication channels and the number of communication channels in use in association with each virtual machine,
The determination as to whether or not to apply the proxy response is made based on a result of comparing the upper limit number of communication channels stored in the application policy management table for each virtual machine with the number of communication channels in use. 5. The inter-virtual machine communication device according to claim 4, wherein the communication device is a virtual machine.
A communication method between virtual machines for performing communication between a plurality of virtual machines on the same virtual machine monitor,
A socket API wrapper library included in at least one of the plurality of virtual machines includes:
Manage communication channels provided by the virtual machine monitor;
An inter-virtual machine communication method for performing a proxy response for inter-virtual machine communication using the communication channel.
The management of the communication channel is as follows:
When communicating between virtual machines, refer to the communication channel management table that associates the communication destination virtual machine with the communication channel or proxy response module in the socket descriptor,
The proxy response is
The inter-virtual machine communication method according to claim 6, which is performed when the proxy response module is associated as a result of referring to the communication channel management table.
A virtual machine running on the virtual machine monitor
Communication channel management means for managing a communication channel provided by the virtual machine monitor;
A socket API wrapper library, which functions as a proxy response means for performing a proxy response for communication between virtual machines using the communication channel.
To a virtual machine running on the virtual machine monitor,
A communication channel management procedure for managing a communication channel provided by the virtual machine monitor;
A computer-readable recording medium storing a program for executing a proxy response procedure for performing a proxy response for communication between virtual machines using the communication channel.