WO2012127598A1

WO2012127598A1 - Synchronization device, synchronization method for synchronization device and synchronization program

Info

Publication number: WO2012127598A1
Application number: PCT/JP2011/056668
Authority: WO
Inventors: 聖金城
Original assignee: 三菱電機株式会社
Priority date: 2011-03-21
Filing date: 2011-03-21
Publication date: 2012-09-27

Abstract

The objective of the present invention is to set a suitable synchronization cycle in order to be able to reduce delay in processing time. A guest OS unit (240) of an operational system physical server (200) carries out predetermined operational processing and performs communication with each terminal device (110). A host OS unit (230) transmits dirty pages of a memory image of the guest OS unit (240) to a standby system physical server (300) for each time that the synchronization cycle elapses. The host OS unit (330) of the standby system physical server (300) receives the dirty pages of the guest OS unit (240), and updates the memory image of the guest OS unit (340) using the received dirty pages. In this regard, the host OS unit (230) updates the synchronization cycle on the basis of operation information (amount of CPU resources, sizes of the dirty pages, and the like) of the guest OS unit (240).

Description

Synchronization device, synchronization method of synchronization device, and synchronization program

The present invention relates to, for example, a synchronization device that synchronizes an active system state and a standby system state, a synchronization method of the synchronization device, and a synchronization program.

A technology that applies virtualization technology, and realizes FT (fault tolerant) in software with a redundant configuration in which the execution state of a virtual machine is synchronized between physical servers (hereinafter referred to as software FT). There is.

In the conventional software FT, as a trigger for synchronization of virtual machines, synchronization control with a fixed period is proposed as in Non-Patent Document 1.
Remus in Non-Patent Document 1 performs synchronization at a fixed period. That is, the value of the synchronization period is specified for Remus before the start of the synchronization process, and thereafter, the synchronization process is performed with the specified synchronization period.

FIG. 8 is a diagram illustrating a configuration example of a system in which Remus is introduced.
In FIG. 8, Xen (reference numerals 003 and 010) is used as a virtual machine monitor.
Remus 004 of the host OS 002 acquires the dirty page 005 (memory difference information) of the synchronization source virtual machine (guest OS 001) as the execution state of the virtual machine. Remus 004 transmits the dirty page 005 to the host OS 007, and Remus 008 of the host OS 007 reflects the dirty page 005 to the synchronization destination virtual machine (guest OS 009). Thereby, the execution state of the virtual machine is synchronized. When Remus 004 acquires the dirty page 005, the operation of the synchronization source virtual machine is temporarily stopped.
In addition, in order to ensure the consistency of input / output (hereinafter referred to as network I / O) between the synchronization source virtual machine and the synchronization destination virtual machine, the buffering mechanism 006 is used to output the network I / O output of the synchronization source virtual machine. Use blocking. At this time, in the synchronization source virtual machine, the blocked network I / O is processed as having been successfully output. The blocked network I / O is sent when the operation of the synchronization source virtual machine is resumed.

JP 2009-080695 A JP 2009-2111517 A

FIG. 9 is a diagram illustrating a processing time when the synchronization processing is not performed.
FIG. 10 and FIG. 11 are diagrams showing a problem when performing the synchronization process.

In FIG. 9, the time at which the “memory access process”, the “data transmission process”, the “data reception process”, and the “idle process” end when the synchronization process is not performed is “t”.
The memory access process is a process for accessing the memory, the data transmission process is a process for transmitting data according to the result of the memory access process, the data reception process is a process for receiving response data for the transmission data, and an idle process Is a process of waiting until response data is received.

As shown in FIG. 10, when a short synchronization cycle is set, a “processing time delay” occurs due to frequent synchronization processing (synchronization commit processing).

As shown in FIG. 11, when a long synchronization cycle is set, the processing time of the memory access process without data transmission is shortened.
However, in order to ensure consistency in data transmission processing between the active system and the standby system, the data transmission processing is blocked until the synchronization processing is completed.
For this reason, in the worst case, the data transmission process is delayed by the amount corresponding to the synchronization period, and a “processing time delay” occurs.

In addition, if memory rewriting occurs frequently in the synchronization source virtual machine after the start of synchronization processing, the dirty page size of the synchronization source virtual machine increases, it takes time to acquire dirty pages, and time required for synchronization processing To increase.

An object of the present invention is to set an appropriate synchronization cycle, for example, and to reduce processing time delay.

The synchronization device according to the present invention executes predetermined operation processing and synchronizes data used in the operation processing with a predetermined standby device.
The synchronization device includes:
A synchronization cycle storage unit that stores a synchronization cycle synchronized with the standby device;
A state data storage unit for storing data used in the operation process as state data;
An operation unit that executes the operation process using the state data stored in the state data storage unit;
An operation stop unit that stops the operation of the operation unit every time the synchronization cycle stored in the synchronization cycle storage unit elapses;
When the operation of the operation unit is stopped by the operation stop unit, a synchronization data transmission unit that transmits at least any data included in the state data to the standby device as synchronization data;
When the synchronization data is transmitted by the synchronization data transmission unit, an operation start unit that starts operation of the operation unit;
A specific time including a first time required to transmit the synchronization data to the standby device, and a specific time at which the ratio of the remaining time after subtracting the first time becomes a predetermined ratio is newly synchronized. A synchronization cycle update unit that calculates the cycle and updates the synchronization cycle stored in the synchronization cycle storage unit with the new synchronization cycle.

According to the present invention, for example, an appropriate synchronization cycle can be set to reduce the delay in operation processing.

1 is a diagram illustrating an example of a configuration of an operation processing system 100 according to Embodiment 1. FIG. FIG. 3 is a functional configuration diagram of an active physical server 200 in the first embodiment. FIG. 3 is a functional configuration diagram of a standby physical server 300 in the first embodiment. FIG. 3 is a schematic diagram of synchronization processing in the first embodiment. 5 is a flowchart showing a synchronization method of the active physical server 200 in the first embodiment. 4 is a flowchart showing a method for dynamically controlling the synchronization period in the first embodiment. 3 is a diagram illustrating an example of hardware resources of an active physical server 200 and a standby physical server 300 according to Embodiment 1. FIG. The figure which shows the structural example of the system which introduced Remus. The figure which shows the processing time when not performing a synchronous process. The figure which shows the subject in the case of performing a synchronous process. The figure which shows the subject in the case of performing a synchronous process. FIG. 6 is a functional configuration diagram of an active physical server 200 according to Embodiment 2. 9 is a flowchart showing a synchronization method of the active physical server 200 in the second embodiment.

Embodiment 1 FIG.
A system configuration for synchronizing the state of the active computer and the state of the standby computer while dynamically updating the synchronization period will be described.

FIG. 1 is a diagram illustrating an example of a configuration of an operation processing system 100 according to the first embodiment.
A configuration example of the operation processing system 100 according to the first embodiment will be described with reference to FIG.

The operation processing system 100 (an example of an information processing system and a synchronization system) includes an active physical server 200 (an example of a synchronization device), a standby physical server 300 (an example of a standby device), and terminal devices 110a-c (an example of a communication device). ).

The active physical server 200 and each of the terminal devices 110a-c are connected to the LAN 102 (an example of a network), communicate via the LAN 102, and perform specific operation processing.

The active physical server 200 and the standby physical server 300 are connected to the LAN 101 (an example of a network), and the memory image of the active physical server 200 is transmitted to the standby physical server 300 at each dynamically changing synchronization period. Synchronize memory images.

A memory image is data stored in a memory.
For example, the memory image means all (or a part) of data stored in a storage area allocated to a virtual machine to be described later.

The standby physical server 300 is connected to the LAN 102, and continues operation processing for each terminal device 110a-c instead of the active physical server 200 when a failure occurs in the active physical server 200.

The operational physical server 200 includes hardware 210, a virtual machine monitor unit 220, a host OS unit 230, and a guest OS unit 240.

The virtual machine monitor unit 220 executes a virtual machine monitor using the hardware 210, constructs (generates) a first virtual machine and a second virtual machine, and performs the first virtual machine and the second virtual machine And control.
The virtual machine monitor is software that emulates a computer and constructs (generates) a virtual computer on an actual computer (physical computer). This virtual computer is called a “virtual machine”.

The host OS unit 230 executes an OS (operating system) as a first virtual machine, and the guest OS unit 240 executes the OS as a second virtual machine.
Hereinafter, the OS of the first virtual machine is referred to as “host OS”, and the OS of the second virtual machine is referred to as “guest OS”.

The host OS unit 230 controls the guest OS unit 240 and transmits the memory image of the guest OS unit 240 to the standby physical server 300.
The guest OS unit 240 executes operation processing. Hereinafter, the guest OS unit 240 is referred to as “synchronization source computer 249”.

Similarly, the standby physical server 300 includes hardware 310, a virtual machine monitor unit 320, a host OS unit 330 (third virtual machine), and a guest OS unit 340 (fourth virtual machine).
Hereinafter, the guest OS unit 340 is referred to as “synchronization destination computer 349”.

The host OS unit 330 controls the guest OS unit 340 (synchronization destination computer 349). For example, the host OS unit 330 receives the memory image of the synchronization source computer 249 (guest OS unit 240) from the active physical server 200, and uses the received memory image to store the memory of the synchronization destination computer 349 (guest OS unit 340). Update the image. As a result, the memory image of the synchronization source computer 249 and the memory image of the synchronization destination computer 349 are synchronized.
When a failure occurs in the synchronization source computer 249, the guest OS unit 240 executes an operation process as the synchronization destination computer 349.

FIG. 2 is a functional configuration diagram of the active physical server 200 in the first embodiment.
The functional configuration of the active physical server 200 in the first embodiment will be described with reference to FIG.

The active physical server 200 (an example of a synchronization device) executes an operation process including a process of transmitting communication data to a specific terminal device 110 (an example of a communication apparatus), and transmits data used in the operation process to a standby system. Synchronizes with the physical server 300 (an example of a standby device).

For example, the operational physical server 200 includes a CPU (Central Processing Unit), a memory, a communication device, a communication buffer, and the like as the hardware 210.

The virtual machine monitor unit 220 controls the host OS unit 230 as a first virtual machine, and controls the guest OS unit 240 as a second virtual machine.
For example, the virtual machine monitor unit 220 allocates CPU usage time to the host OS unit 230 and the guest OS unit 240 based on a control table stored in advance in the memory. The CPU usage time is the time per unit time in which the CPU is used (also referred to as CPU usage rate).
Further, the virtual machine monitor unit 220 allocates memory storage areas to the host OS unit 230 and the guest OS unit 240 based on the control table.
The control table is data in which virtual machine control information is set. For example, the CPU usage time ratio and the storage area address are set in the control table so as to correspond to the host OS unit 230 and the guest OS unit 240, respectively.

Hereinafter, a storage area allocated to the host OS unit 230 is referred to as a “host OS area”, and a storage area allocated to the guest OS unit 240 is referred to as a “guest OS area (an example of a state data storage unit)”.
A storage area used by the virtual machine monitor unit 220 is referred to as a “monitor area (an example of an operation information storage unit)”.

For example, the monitor area stores a synchronous communication speed (Wc described later), a data communication speed (Wd described later), an update rate (Rr described later), and a generation rate (Gr described later).
The synchronous communication speed is a communication speed when communicating with the standby physical server 300.
The data communication speed is a communication speed when communicating with the terminal device 110.
The update rate represents the size of data updated per unit time among data included in the memory image (an example of state data) of the guest OS area.
The generation rate represents the size of communication data generated per unit time.

The guest OS unit 240 is a synchronization source computer 249 provided with an operation unit 241.

The operation unit 241 uses the memory image of the guest OS area (an example of state data) to execute operation processing.

The host OS unit 230 includes an operation information acquisition unit 231, a synchronization cycle calculation unit 232, a synchronization cycle storage unit 233, a synchronization start instruction unit 234, and a synchronization engine unit 235.

The synchronization cycle storage unit 233 (an example of the synchronization cycle storage unit) is a part of the host OS area.
The synchronization period storage unit 233 stores a synchronization period for synchronizing the synchronization source computer 249 (guest OS unit 240) and the synchronization destination computer 349 (guest OS unit 340).

The synchronization engine unit 235 (an example of an operation stop unit, a synchronization data transmission unit, and an operation start unit) performs the following synchronization processing.

The synchronization engine unit 235 stops the operation of the operation unit 241 every time the synchronization period stored in the synchronization period storage unit 233 elapses (operation stop process).
When the operation of the operation unit 241 is stopped, the synchronization engine unit 235 transmits at least one data included in the memory image to the standby physical server 300 as synchronization data (synchronous data transmission processing). For example, the synchronization engine unit 235 transmits, to the standby physical server 300, the data (dirty page to be described later) updated when the operation unit 241 is operating among the data included in the memory image as the synchronization data. Further, for example, the synchronization engine unit 235 transmits the entire memory image to the standby physical server 300 as synchronization data.
The synchronization engine unit 235 starts the operation of the operation unit 241 when the synchronization data is transmitted (operation start process).

The synchronization cycle calculation unit 232 (an example of the synchronization cycle update unit) updates the synchronization cycle (Δt described later) as follows (synchronization cycle update processing).
The synchronization period calculation unit 232 newly sets a specific time at which a ratio (Pvm, which will be described later) of a remaining time obtained by subtracting a synchronization processing time (to be described later) and an I / O processing delay (Tio, which will be described later) becomes a predetermined ratio. Calculated as the synchronization period.
The synchronization cycle calculation unit 232 updates the synchronization cycle stored in the synchronization cycle storage unit 233 with a new synchronization cycle.

The synchronization processing time (an example of the first time) is a time required for transmitting the synchronization data to the standby physical server 300.
The I / O processing delay (an example of the second time) is a time required for transmitting communication data to the terminal device 110.

For example, the synchronization cycle calculation unit 232 calculates, as a new synchronization time, a specific time at which the ratio of the remaining time obtained by subtracting the synchronization processing time and the I / O processing delay is maximized.

For example, the synchronization processing time is a time including a dirty page transmission time (synchronization data transmission time) obtained by dividing the data size of the synchronization data by the synchronization communication speed (Wc).
The I / O processing delay is a time including a time (communication data transmission time) obtained by dividing the data size of the communication data by the data communication speed (Wd).

For example, the data size of the synchronous data is a data size obtained by multiplying the update rate (Rr) by a specific time.
The data size of the communication data is a data size obtained by multiplying the generation rate (Gr) by a specific time.

The operation information acquisition unit 231 acquires operation information of the guest OS unit 240 from the monitor area via the virtual machine monitor unit 220.
For example, the synchronous communication speed (Wc), the data communication speed (Wd), the update rate (Rr), and the generation rate (Gr) are examples of operation information.

The synchronization start instruction unit 234 instructs the synchronization engine unit 235 to execute the synchronization process every time the synchronization period elapses.

FIG. 3 is a functional configuration diagram of the standby physical server 300 according to the first embodiment.
The functional configuration of the standby physical server 300 in the first embodiment will be described with reference to FIG.

The standby physical server 300 (an example of a standby device) includes hardware 310, a virtual machine monitor unit 320, a host OS unit 330, and a guest OS unit 340, like the active physical server 200 (see FIG. 2).

For example, the standby physical server 300 includes a CPU and a memory as the hardware 310.

The virtual machine monitor unit 320 controls the host OS unit 330 as a third virtual machine, and controls the guest OS unit 340 as a fourth virtual machine.

Hereinafter, a storage area allocated to the host OS unit 330 is referred to as a “host OS area”, a storage area allocated to the guest OS unit 340 is referred to as a “guest OS area”, and a storage area used by the virtual machine monitor unit 320 is referred to as a “host OS area”. This is called “monitor area”.

The guest OS unit 340 is a synchronization destination computer 349 including an operation unit 341.
The operation unit 341 executes operation processing instead of the operation unit 241 when a failure occurs in the operation unit 241 (see FIG. 2) of the synchronization source computer 249.

The host OS unit 330 includes a synchronization server unit 331.
The synchronization server unit 331 receives the synchronization data transmitted from the active physical server 200, and updates the memory image of the guest OS unit 340 (the memory image of the guest OS area) using the received synchronization data.

Next, a synchronization method executed for each synchronization cycle and a synchronization cycle dynamic control method for updating the synchronization cycle will be described.

FIG. 4 is a schematic diagram of synchronization processing in the first embodiment.
An overview of the synchronization processing in the first embodiment will be described with reference to FIG.

In the active physical server 200, the synchronization engine unit 235 of the host OS unit 230 temporarily stops the synchronization source computer 249 (guest OS unit 240) for each synchronization period.
The synchronization period is the time from when the synchronization source computer 249 is temporarily stopped until the next temporary stop.

The synchronization engine unit 235 suspends the synchronization source computer 249 and then acquires an updated page among the pages included in the memory image of the synchronization source computer 249. Hereinafter, the updated page is referred to as a “dirty page”.
The synchronization engine unit 235 transmits the acquired dirty page to the standby physical server 300.
In the standby physical server 300, the synchronization server unit 331 of the host OS unit 330 receives the dirty page, and uses the received dirty page to correspond to the dirty page in the memory image of the synchronization target computer 349 (guest OS unit 340). Refresh the page.

The synchronization engine unit 235 restarts the synchronization source computer 249 after transmitting the dirty page.
However, the synchronization engine unit 235 buffers the transmission data output from the synchronization source computer 249 when the synchronization source computer 249 is restarted. The transmission data of the synchronization source computer 249 is accumulated in the communication buffer by buffering.
Hereinafter, buffering the transmission data of the synchronization source computer 249 is referred to as “I / O blocking”.

The restarted synchronization source computer 249 executes operation processing, generates transmission data to be transmitted to the terminal device 110, and outputs the generated transmission data to the communication buffer.
The communication device of the hardware 210 transmits the transmission data stored in the communication buffer in the previous synchronization cycle, and stores the transmission data generated in the current synchronization cycle in the communication buffer.

FIG. 5 is a flowchart showing a synchronization method of the active physical server 200 in the first embodiment.
A method of synchronizing the active physical server 200 in the first embodiment will be described with reference to FIG.

First, an overview of the synchronization method of the active physical server 200 will be described.

When the synchronization period has elapsed (S110), the synchronization engine unit 235 executes the following process.

The synchronization engine unit 235 starts synchronization processing (S120), stops the I / O blocking of the synchronization source computer 249 (S121), and temporarily stops the synchronization source computer 249 (S122).
The synchronization engine unit 235 acquires the dirty page of the synchronization source computer 249 (S130), and transmits the acquired dirty page to the synchronization destination computer 349 (S131).
The synchronization engine unit 235 restarts the synchronization source computer 249 (S140), starts I / O blocking of the synchronization source computer 249 (S141), and ends the synchronization process (S142).

The synchronization start instruction unit 234 refers to the synchronization period stored in the synchronization period storage unit 233 (S150), and starts a synchronization period timer that notifies the elapse of the synchronization period (S151).

Next, details of the synchronization method of the active physical server 200 will be described.

In S110, the synchronization start instruction unit 234 of the host OS unit 230 waits until the synchronization cycle timer started in the previous S151 times out.
When the synchronization start instruction unit 234 detects a timeout of the synchronization cycle timer, the synchronization start instruction unit 234 instructs the synchronization engine unit 235 to execute the synchronization process. For example, the synchronization start instruction unit 234 calls the synchronization engine unit 235.
After instructing the synchronization engine unit 235 to execute the synchronization process, the process proceeds to S120.

In S120, the synchronization engine unit 235 of the host OS unit 230 receives the execution instruction of the synchronization process from the synchronization start instruction unit 234, and starts executing the synchronization process (S121-S141) described below.
Further, the synchronization engine unit 235 starts measuring the processing time required for the synchronization processing.
It progresses to S121 after S120.

In S121, the synchronization engine unit 235 stops the I / O blocking of the synchronization source computer 249 (guest OS unit 240) started in the previous S141 as the “transmission control unit”.
At this time, the synchronization engine unit 235 requests the virtual machine monitor unit 220 to stop the I / O blocking, and the virtual machine monitor unit 220 controls the communication buffer allocated to the communication device or the synchronization source computer 249 to control the synchronization source computer 249. Stop I / O blocking.
The I / O blocking of the synchronization source computer 249 is a process or function for storing the transmission data generated by the synchronization source computer 249 in the communication buffer without transmitting it.
It progresses to S122 after S121.

In S122, the synchronization engine unit 235 temporarily stops the synchronization source computer 249.
At this time, the synchronization engine unit 235 requests the virtual machine monitor unit 220 to temporarily stop the synchronization source computer 249, and the virtual machine monitor unit 220 temporarily stops the synchronization source computer 249.
For example, the virtual machine monitor unit 220 temporarily stops the allocation of the CPU usage time to the synchronization source computer 249.
Thereafter, the execution of the operation process by the operation unit 241 is stopped until the synchronization source computer 249 is restarted.
It progresses to S130 after S122.

The order of S121 and S122 may be reversed. Further, S121 may be executed in S140.

In S130, the synchronization engine unit 235 acquires the dirty page of the synchronization source computer 249.
At this time, the synchronization engine unit 235 requests the virtual machine monitor unit 220 to acquire a dirty page, the virtual machine monitor unit 220 acquires the dirty page from the memory image of the synchronization source computer 249, and acquires the acquired dirty page as the synchronization engine unit. In response to H.235.
The memory image of the synchronization source computer 249 is data stored in the storage area allocated to the synchronization source computer 249. The memory image includes a plurality of pages.
A dirty page is a page that has been updated since the last acquisition of the dirty page among the pages included in the memory image. Dirty pages are also called update pages.
The dirty page is managed by the OS of the active physical server 200 or the virtual machine monitor unit 220 (an example of the update data management unit) using a management table for managing the storage area.
It progresses to S131 after S130.

In S131, the synchronization engine unit 235 transmits the dirty page of the synchronization source computer 249 to the synchronization destination computer 349.
At this time, the synchronization engine unit 235 generates data (communication packet) including a dirty page and is transmitted to the standby physical server 300 as synchronization data, and stores the data in the communication buffer allocated to the host OS unit 230. Set the synchronization data.
The communication device transmits the synchronization data set in the communication buffer.
The synchronization server unit 331 of the standby physical server 300 receives the synchronization data, and acquires the dirty page of the synchronization source computer 249 from the received synchronization data. The synchronization server unit 331 updates the page corresponding to the acquired dirty page among the pages included in the memory image of the synchronization destination computer 349 (guest OS unit 340) using the acquired dirty page. At this time, the synchronization server unit 331 requests the virtual machine monitor unit 320 to update the memory image of the synchronization destination computer 349, and the virtual machine monitor unit 320 uses the dirty page of the synchronization source computer 249 to store the memory image of the synchronization destination computer 349. Update.
After S131, the process proceeds to S140.

In S140, the synchronization engine unit 235 requests the virtual machine monitor unit 220 to restart the synchronization source computer 249, and the virtual machine monitor unit 220 restarts the synchronization source computer 249.
At this time, the virtual machine monitor unit 220 controls the communication buffer of the communication device or the synchronization source computer 249 to transmit the transmission data buffered in the previous synchronization cycle to the terminal device 110. However, the transmission data buffered in the previous synchronization cycle may be transmitted by the communication device when S121 is executed.
Thereafter, the virtual machine monitor unit 220 resumes the allocation of the CPU usage time to the synchronization source computer 249.
Thereafter, operation processing is executed by the operation unit 241 until the synchronization source computer 249 is temporarily stopped.
It progresses to S141 after S140.

In S141, the synchronization engine unit 235 starts I / O blocking of the guest OS unit 240 as a “transmission control unit”.
At this time, the synchronization engine unit 235 requests the virtual machine monitor unit 220 to start I / O blocking, and the virtual machine monitor unit 220 controls the communication buffer allocated to the communication device or the synchronization source computer 249 to control the synchronization source computer 249. Start I / O blocking.
Thereafter, transmission data generated by the synchronization source computer 249 is accumulated in the communication buffer until the I / O blocking is stopped.
It progresses to S142 after S141.

The order of S140 and S141 may be reversed.

In S142, the synchronization engine unit 235 ends the execution of the synchronization process (S121-S141).
Also, the synchronization engine unit 235 ends the measurement of the processing time required for the synchronization process.
It progresses to S150 after S142.

In S <b> 150, the synchronization start instruction unit 234 refers to the synchronization cycle storage unit 233 and acquires the synchronization cycle stored in the synchronization cycle storage unit 233.
After S150, the process proceeds to S151.

In S151, the synchronization start instruction unit 234 subtracts the processing time of the synchronization process measured from S120 to S142 from the synchronization period. Hereinafter, the time obtained by subtracting the processing time of the synchronization process from the synchronization period is referred to as “the remaining time of the synchronization period”.
The synchronization start instruction unit 234 sets the remaining time of the synchronization period in the timer, and starts the timer in which the remaining time of the synchronization period is set. In the embodiment, the timer in which the remaining time of the synchronization period is set is referred to as “synchronization period timer”.
The synchronization cycle timer is a function that times out when the remaining time of the synchronization cycle elapses and notifies the synchronization start instruction unit 234 of the timeout.
After S151, the process returns to S110.

That is, the synchronization cycle timer times out every time the synchronization cycle elapses (S110), and the synchronization processing (S120-S142) is executed for each synchronization cycle.

FIG. 6 is a flowchart illustrating a synchronization cycle dynamic control method according to the first embodiment.
A method for dynamically controlling the synchronization period in the first embodiment will be described with reference to FIG.

First, the outline of the dynamic control method of the synchronization cycle will be described.

The operation information acquisition unit 231 acquires operation information of the synchronization source computer 249 (S210).
The synchronization cycle calculation unit 232 calculates a new synchronization cycle based on the operation information of the synchronization source computer 249 (S220).
The synchronization cycle calculation unit 232 updates the synchronization cycle stored in the synchronization cycle storage unit 233 to a new synchronization cycle (S230).
S210 to S230 are executed at every predetermined update timing (S240).

Next, the details of the synchronous cycle dynamic control method will be described.

In S210, the operation information acquisition unit 231 acquires operation information of the synchronization source computer 249 (guest OS unit 240).
At this time, the operation information acquisition unit 231 requests the virtual machine monitor unit 220 for operation information of the synchronization source computer 249. The virtual machine monitor unit 220 acquires the operation information of the synchronization source computer 249 from the memory of the hardware 210, and returns the acquired operation information of the synchronization source computer 249 to the operation information acquisition unit 231.

For example, the operation information acquisition unit 231 includes “CPU resource amount” “Dirty page generation rate” “Dirty page acquisition rate” “Synchronization preparation time” “Dirty page communication bandwidth” “Transmission data generation rate” “Transmission data” ”Is acquired as operation information of the synchronization source computer 249.
For example, a table in which operation information of the synchronization source computer 249 is set is referred to as an “operation information table”.

“CPU resource amount” is the percentage of CPU usage time allocated to the synchronization source computer 249. For example, when 30% of time is allocated as the CPU usage time of the synchronization source computer 249, the CPU resource amount is “0.3 (= 30/100)”.

“Dirty page generation rate” is the size of a page updated per unit time among pages included in the memory image of the synchronization source computer 249. For example, the unit is “KB / s” or “page / s” (KB: kilobyte, s: second).

“Dirty page acquisition rate” is the time required to acquire a unit amount of dirty pages from the memory image of the synchronization source computer 249. For example, the unit is “s / KB” or “s / page”.

“Synchronization preparation time” is the total time of the time required to temporarily stop the synchronization source computer 249 and the time required to restart the synchronization source computer 249. For example, the unit is “s”.

The “dirty page communication bandwidth” is the communication speed of the LAN 101 used when a dirty page (synchronized data) is transmitted from the active physical server 200 to the standby physical server 300. For example, the unit is “KB / s”.

“The transmission data generation rate” is the size of transmission data generated by the synchronization source computer 249 per unit time. For example, the unit is “KB / s”.

“Communication band of transmission data” is a communication speed of the LAN 102 used when transmission data is transmitted from the active physical server 200 to the terminal device 110. For example, the unit is “KB / s”.

After S210, the process proceeds to S220.

In S220, the synchronization cycle calculation unit 232 calculates a new synchronization cycle based on the operation information of the synchronization source computer 249 acquired in S210.

At this time, the synchronization cycle calculation unit 232 calculates, as a new synchronization cycle, a synchronization cycle in which the effective operating rate of the synchronization source computer 249 is maximized. However, the synchronization cycle calculation unit 232 may calculate a synchronization cycle other than the synchronization cycle that maximizes the execution availability as a new synchronization cycle. For example, the synchronization cycle calculation unit 232 may calculate a synchronization cycle in which the execution operation rate becomes a predetermined operation rate as a new synchronization cycle.
The effective operating rate of the synchronization source computer 249 is the ratio of the actual VM operation time to the synchronization period, and the actual VM operation time is the time for the operation unit 241 to actually execute the operation processing (with an I / O processing delay) VM operating time).
Equation (1) representing the effective operating rate of the synchronization source computer 249 is shown below.

Next, as equations for calculating the effective operating rate of the synchronization source computer 249, equation (2) representing the synchronization processing time, equation (3) representing the VM operation time, and equation representing the I / O processing delay ( 4) and Expression (5) representing the actual VM operation time are shown.

In the above equation (2), the “synchronization processing time” corresponds to the time required for the synchronization processing (S120 to S142 in FIG. 5).
“Dirty page acquisition time” is the time required to acquire a dirty page.
“Dirty page size” (synchronization data size) is the size of a page (dirty page) that is updated per synchronization period among pages included in the memory image of the synchronization source computer 249.
The “dirty page transmission time” is the time from when the active physical server 200 transmits a dirty page until the standby physical server 300 receives the dirty page.
However, any one of “dirty page acquisition time”, “dirty page transmission time”, and “synchronization preparation time”, or any two total time may be used as the synchronization processing time.

In the above equation (3), “VM operation time” (operation unit operation time) is the time during which the synchronization source computer 249 (operation unit 241) operates in the synchronization period.

In the above formula (4), “I / O processing delay” (communication data transmission time) is the time required for transmission data transmission.
“Transmission data size” (communication data size) is the size of transmission data generated per synchronization cycle.

However, “I / O processing delay” may be ignored. That is, the effective operating rate of the synchronization source computer 249 may be expressed by the following equation (6).

After S220, the process proceeds to S230.

In S230, the synchronization cycle calculation unit 232 updates the current synchronization cycle stored in the synchronization cycle storage unit 233 with the new synchronization cycle calculated in S220.
It progresses to S240 after S230.

In S240, the synchronization cycle calculation unit 232 determines whether or not a predetermined update timing has come.
For example, the synchronization cycle calculation unit 232 activates an update timer that sets a predetermined update time, and determines that the update timing has come when the update timer times out. When the update timer times out, the synchronization cycle calculation unit 232 newly starts the update timer.
Further, the synchronization cycle calculation unit 232 may determine that the update timing has come each time the synchronization cycle timer (see S151 in FIG. 5) times out by a predetermined number of synchronizations.
If the update timing is reached, the process returns to S210.

The operation information of the synchronization source computer 249 may be a fixed value or a variable value. When the operation information of the synchronization source computer 249 is set to a variable value, the virtual machine monitor unit 220 manages the operation information of the synchronization source computer 249 as an “operation information management unit”.
However, the host OS unit 230 may have the function of “operation information management unit”.

The virtual machine monitor unit 220 stores the past synchronization cycle and the past dirty page size in the memory, and generates a new “dirty page generation rate based on the past synchronization cycle and the past dirty page size stored in the memory. May be calculated.
For example, the virtual machine monitor unit 220 calculates a new “dirty page generation rate” based on the previous synchronization cycle and the previous dirty page size.
Further, the virtual machine monitor unit 220 stores the past “dirty page generation rate” in the memory, and calculates the average value of the past “dirty page generation rate” as a new “dirty page generation rate”. Also good.

The virtual machine monitor unit 220 stores the past synchronization cycle and the past transmission data size in the memory, and generates a new “transmission data” based on the past synchronization cycle and the past transmission data size stored in the memory. The “generation rate” may be calculated.
For example, the virtual machine monitor unit 220 calculates a new “transmission data generation rate” based on the previous synchronization period and the previous transmission data size.
Further, the virtual machine monitor unit 220 stores the past “transmission data generation rate” in the memory, and calculates an average value of the past “transmission data generation rate” as a new “transmission data generation rate”. Also good.

FIG. 7 is a diagram illustrating an example of hardware resources of the active physical server 200 and the standby physical server 300 according to the first embodiment.
In FIG. 7, the active physical server 200 and the standby physical server 300 include a CPU 901 (Central Processing Unit). The CPU 901 is connected to a ROM 903, a RAM 904, a communication board 905 (an example of a communication device), a display device 911, a keyboard 912, a mouse 913, a drive device 914, and a magnetic disk device 920 via a bus 902, and these hardware devices are connected. Control. The drive device 914 is a device that reads and writes a storage medium such as an FD (Flexible Disk Drive), a CD (Compact Disc), and a DVD (Digital Versatile Disc).

The communication board 905 is wired or wirelessly connected to a communication network such as a LAN (Local Area Network), the Internet, or a telephone line.

The magnetic disk device 920 stores an OS 921 (operating system), a program group 922, and a file group 923.

The program group 922 includes a program for executing a function described as “unit” in the embodiment. The program is read and executed by the CPU 901. That is, the program causes the computer to function as “˜part”, and causes the computer to execute the procedures and methods of “˜part”.

The file group 923 includes various data (input, output, determination result, calculation result, processing result, etc.) used in “˜part” described in the embodiment.

In the embodiment, arrows included in the configuration diagrams and flowcharts mainly indicate input and output of data and signals.

In the embodiment, what is described as “to part” may be “to circuit”, “to apparatus”, and “to device”, and “to step”, “to procedure”, and “to processing”. May be. That is, what is described as “˜unit” may be implemented by any of firmware, software, hardware, or a combination thereof.

In the first embodiment, for example, the following synchronous cycle dynamic control method has been described.
The active physical server 200 includes a synchronization source virtual machine (synchronization source computer 249) on which the first guest OS operates and a virtual machine on which a first host OS (host OS unit 230) that manages the synchronization source virtual machine operates. With.
The standby physical server 300 includes a synchronization destination virtual machine (synchronization destination computer 349) on which the second guest OS operates and a virtual machine on which a second host OS (host OS unit 330) that manages the synchronization destination virtual machine operates. With.
The first host OS synchronizes the execution state of the synchronization source virtual machine operating on the active physical server 200 and the execution state of the synchronization destination virtual machine on the standby physical server 300 by periodic control.
The first host OS includes an operation information acquisition unit 231, a synchronization cycle calculation unit 232, and a synchronization cycle storage unit 233, and dynamically changes the synchronization cycle.
The operation information acquisition unit 231 acquires operation information (CPU resource amount, dirty page size, data transmission amount) of the synchronization source virtual machine during the execution of the synchronization process.
The synchronization cycle calculation unit 232 calculates the synchronization cycle using the motion information acquired by the motion information acquisition unit 231 and the evaluation formula.
The synchronization cycle storage unit 233 is a shared data area that stores the synchronization cycle calculated by the synchronization cycle calculation unit 232.

Generated by synchronous commit processing by calculating the synchronization period using the operation information (CPU resource amount, dirty page size, data transmission amount) of the synchronization source virtual machine and the evaluation formula, and dynamically controlling the synchronization period Overhead (operation processing delay time) can be reduced.

In the first embodiment, not the dirty page but the entire memory image may be transmitted from the active physical server 200 to the standby physical server 300.
The host OS unit (reference numerals 230 and 330) and the guest OS part (reference numerals 240 and 340) may not be virtual machines.
A plurality of CPUs may be provided, a specific CPU may be assigned to the host OS unit, and another CPU may be assigned to the guest OS unit.

Embodiment 2. FIG.
A description will be given of a mode in which the CPU usage time allocated to the guest OS unit 240 is allocated to the host OS unit 230 when executing the synchronization process.
Hereinafter, items different from the first embodiment will be mainly described. Matters whose description is omitted are the same as those in the first embodiment.

FIG. 12 is a functional configuration diagram of the active physical server 200 according to the second embodiment.
The functional configuration of the active physical server 200 in the second embodiment will be described with reference to FIG.

The host OS unit 230 of the active physical server 200 includes a CPU allocation unit 236 in addition to the configuration described in the first embodiment (see FIG. 2).

The CPU allocation unit 236 allocates the CPU usage time allocated to the operation unit 241 to the synchronization engine unit 235 when the operation of the operation unit 241 is stopped.

FIG. 13 is a flowchart showing a synchronization method of the active physical server 200 in the second embodiment.
A method of synchronizing the active physical server 200 in the second embodiment will be described with reference to FIG.

In addition to the processing described in the first embodiment (see FIG. 5), S123 and S132 are executed.
Hereinafter, S123 and S132 will be described.

In S123, the synchronization source computer 249 is in an idle state in which the CPU usage time is not allocated because it is temporarily stopped in S122.
Therefore, the CPU allocation unit 236 requests the virtual machine monitor unit 220 to add the allocation, and the virtual machine monitor unit 220 additionally allocates the CPU usage time allocated to the synchronization source computer 249 to the host OS unit 230.

After S123, the synchronization engine unit 235 executes S130 and S131 using the CPU based on the CPU usage time originally allocated to the host OS unit 230 and the CPU usage time allocated additionally.
Thereby, the processing time of S130 and S131 can be shortened.
After S131, the process proceeds to S132.

In S <b> 132, the synchronization source computer 249 requests the virtual machine monitor unit 220 to cancel the allocation, and the virtual machine monitor unit 220 reallocates the CPU usage time added and allocated to the host OS unit 230 to the synchronization source computer 249 again.
It progresses to S140 after S132.

The synchronous cycle dynamic control method is the same as that of the first embodiment (see FIG. 6).
However, since the CPU usage time allocated to the synchronization source computer 249 is allocated to the host OS unit 230, the “CPU resource amount” (see equation (2)) for calculating the synchronization processing time increases.
As a result, the synchronization processing time is shortened compared to the first embodiment, the effective operating rate of the synchronization source computer 249 is increased, and the synchronization cycle is changed.

In the second embodiment, for example, the following synchronous cycle dynamic control method has been described.
The first host OS (host OS unit 230) of the active physical server 200 includes a CPU allocation unit 236.
When the CPU allocation unit 236 acquires the execution state of the synchronization source virtual machine (synchronization source computer 249) on which the first guest OS operates, the unused CPU core (or CPU usage) of the active physical server 200 is acquired. Time) is allocated to the first host OS.

By allocating unused CPU resources (CPU usage time) to the management host (first host OS) that is performing the synchronization processing, the processing time of the synchronization processing can be shortened.

001 guest OS, 002 host OS, 003 Xen, 004 Remus, 005 dirty page, 006 buffering mechanism, 007 host OS, 008 Remus, 009 guest OS, 010 Xen, 100 operation processing system, 101, 102 LAN, 110 terminal device , 200 operational physical server, 210 hardware, 220 virtual machine monitor unit, 230 host OS unit, 231 operation information acquisition unit, 232 synchronization cycle calculation unit, 233 synchronization cycle storage unit, 234 synchronization start instruction unit, 235 synchronization engine unit 236 CPU allocation unit, 240 guest OS unit, 241 operation unit, 249 synchronization source computer, 300 standby physical server, 310 hardware, 320 virtual machine monitor unit, 330 host OS section, 331 synchronization server section, 340 guest OS section, 341 operation section, 349 synchronization destination computer, 901 CPU, 902 bus, 903 ROM, 904 RAM, 905 communication board, 911 display device, 912 keyboard, 913 mouse, 914 drive Device, 920 magnetic disk device, 921 OS, 922 program group, 923 file group.

Claims

In a synchronization device that executes predetermined operation processing and synchronizes data used in the operation processing with a predetermined standby device,
A synchronization cycle storage unit that stores a synchronization cycle synchronized with the standby device;
A state data storage unit for storing data used in the operation process as state data;
An operation unit that executes the operation process using the state data stored in the state data storage unit;
An operation stop unit that stops the operation of the operation unit every time the synchronization cycle stored in the synchronization cycle storage unit elapses;
When the operation of the operation unit is stopped by the operation stop unit, a synchronization data transmission unit that transmits at least any data included in the state data to the standby device as synchronization data;
When the synchronization data is transmitted by the synchronization data transmission unit, an operation start unit that starts operation of the operation unit;
A specific time including a first time required to transmit the synchronization data to the standby device, and a specific time at which the ratio of the remaining time after subtracting the first time becomes a predetermined ratio is newly synchronized. A synchronization apparatus comprising: a synchronization cycle update unit that calculates a cycle and updates a synchronization cycle stored in the synchronization cycle storage unit with the new synchronization cycle.
The synchronization apparatus according to claim 1, wherein the synchronization period update unit calculates a specific time at which the ratio of the remaining time is maximized as the new synchronization time.
The synchronization device further includes:
An operation information storage unit that stores a communication speed when communicating with the standby device as a synchronous communication speed;
3. The synchronization device according to claim 2, wherein the first time is a time including a synchronous data transmission time obtained by dividing a data size of the synchronous data by the synchronous communication speed.
The operation information storage unit stores, as an update rate, the size of data updated in unit time among the data included in the state data,
4. The synchronization device according to claim 3, wherein the data size of the synchronization data is a data size obtained by multiplying the update rate by the specific time.
The operation process includes a process of transmitting communication data to a specific communication device,
The synchronization period update unit sets the specific time at which the ratio of the remaining time obtained by subtracting the first time and the second time required to transmit the communication data to the communication device is the predetermined ratio. 5. The synchronization device according to claim 4, wherein the synchronization device is calculated as a stable synchronization cycle.
The operation information storage unit stores a communication speed when communicating with the communication device as a data communication speed,
6. The synchronization apparatus according to claim 5, wherein the second time is a time including a communication data transmission time obtained by dividing the data size of the communication data by the data communication speed.
The operation information storage unit stores the size of communication data generated per unit time as a generation rate,
The synchronization apparatus according to claim 6, wherein the data size of the communication data is a data size obtained by multiplying the generation rate by the specific time.
The synchronization device further includes:
2. The CPU allocation unit for allocating a usage time of a CPU (Central Processing Unit) allocated to the operation unit to the synchronous data transmission unit when the operation of the operation unit is stopped. Synchronization device.
The synchronization device further includes a virtual machine monitor unit that generates a first virtual machine and a second virtual machine,
The operation unit operates on the first virtual machine,
The synchronization apparatus according to claim 1, wherein the operation stop unit, the synchronization data transmission unit, the operation start unit, and the synchronization cycle update unit operate in the second virtual machine.
A synchronization device that executes predetermined operation processing and synchronizes data used in the operation processing with a predetermined standby device, the synchronization cycle storage unit storing a synchronization cycle synchronized with the standby device, and the operation processing In a synchronization method of a synchronization device comprising a state data storage unit that stores data used in the state data as state data,
The operation unit executes the operation process using the state data stored in the state data storage unit,
The operation stop unit stops the operation of the operation unit every time the synchronization cycle stored in the synchronization cycle storage unit elapses,
When the operation of the operation unit is stopped by the operation stop unit, the synchronization data transmission unit transmits at least any data included in the state data to the standby device as synchronization data,
When the operation start unit transmits the synchronization data by the synchronization data transmission unit, the operation start unit starts the operation of the operation unit,
The specific period including the first time required for the synchronization period updating unit to transmit the synchronization data to the standby device, and the ratio of the remaining time after subtracting the first time is a predetermined ratio A synchronization method for a synchronization apparatus, comprising: calculating a specific time as a new synchronization period, and updating the synchronization period stored in the synchronization period storage unit with the new synchronization period.
A synchronization device that executes predetermined operation processing and synchronizes data used in the operation processing with a predetermined standby device, the synchronization cycle storage unit storing a synchronization cycle synchronized with the standby device, and the operation processing In a synchronization program for functioning a synchronization device comprising a state data storage unit for storing data used in the state data as state data,
An operation unit that executes the operation process using the state data stored in the state data storage unit;
An operation stop unit that stops the operation of the operation unit every time the synchronization cycle stored in the synchronization cycle storage unit elapses;
When the operation of the operation unit is stopped by the operation stop unit, a synchronization data transmission unit that transmits at least any data included in the state data to the standby device as synchronization data;
When the synchronization data is transmitted by the synchronization data transmission unit, an operation start unit that starts operation of the operation unit;
A specific time including a first time required to transmit the synchronization data to the standby device, and a specific time at which the ratio of the remaining time after subtracting the first time becomes a predetermined ratio is newly synchronized. A synchronization program that causes the synchronization device to function as a synchronization cycle update unit that calculates a cycle and updates a synchronization cycle stored in the synchronization cycle storage unit with the new synchronization cycle.