JP4156470B2 - Node transfer device, node alternative device and program thereof - Google Patents

Description

  The present invention provides a node transfer device that is a transfer source, a node transfer device that has the same hardware configuration as the node transfer device, and is based on a copy of a memory area of the node transfer device that is written to an external storage device for the node transfer device. The present invention relates to a node alternative device that takes over the process of a transfer device and a program thereof.

Servers (PC clusters) with a plurality of commodity PCs can realize a server function having high availability and flexible scalability at low cost, and have been actively developed. In a PC cluster, it is necessary to perform maintenance by stopping a node for addition or change of hardware or software. However, as a whole service, it is necessary to minimize the influence of the node stop.
Therefore, in order to increase the availability of the node, a node (substitute node) as an alternative to the active node (main node) is prepared. By moving the processing of the main node to an alternative node (transfer), it is possible to continue the running service even when the main node is stopped or under maintenance. There are the following technologies related to node processing transfer.

In a PC cluster, when a specific node environment is migrated to another node, a roll forward method is common. The roll-forward method is a method in which a program of a process executed on the migration source node is re-executed on the migration destination node and restored to the same process state as before the migration.
There is a virtual machine method as a method for migrating a stand-alone node environment that is not a cluster to another node. The virtual computer method is a method of constructing a virtual node environment using software and transferring the virtual node to a virtual node environment of another node.
Further, as a technique for resuming a suspended process in the same node, Patent Document 1 discloses a plurality of applications that restart an application existing in another terminal on the network that was operating when the power is turned off in the same state as when the power is turned off. An invention related to a terminal application activation method is disclosed.
JP 2000-322243 A

However, the conventional techniques described above have the following problems.
First, in the roll forward method, the internal state of the process being executed at the migration source node is reset at the migration destination node. For this reason, it is necessary to roll forward (retrace to the latest state) from an intermediate state of the process at the application level. Since this is a process that depends on the application, there is a problem in that the applications that can be handled are limited.
Also, in an OS environment without a resource reservation function, the resource ID (process identifier or socket port number) assigned to the application restarted at the time of migration differs before and after the migration, so complete transparency cannot be guaranteed. There is a problem.
Furthermore, at the application level, the same one is started, but the node itself is different from the one before the migration, so the IP address used in the network interface changes. Therefore, address translation at the network layer is necessary. However, when such a conversion is performed, it affects the inside and outside of the migration source note and the migration destination node (such as the routing table of the network equipment outside the same network). Continuing to do so has the problem of complicated management.

In the virtual machine method, the target OS operates on the virtual machine, but there is an overhead of virtual machine software. For this reason, there is a problem that the CPU is consumed by the virtual machine software even during normal use, and 100% performance cannot be exhibited.
In addition, there is a problem that only devices supported by the virtual machine software can be used on the virtual machine.

  The present invention has been made in view of such circumstances, and an object of the present invention is to provide a node transfer device, a node replacement device, and a program thereof that can take over the environment of the transfer source node as it is to the transfer destination node. There is.

The present invention has been made to solve the above-described problems. The present invention comprises a node transport device and the same hardware configuration as the node transport device, and is written to an external storage device for the node transport device. A node transport device in a computer system comprising a node replacement device that takes over the process of the node transport device based on a copy of the memory area of the node transport device to be executed, and executed in the node transport device other than the node transport process In the suspended state in which all processes in the process are suspended , the control unit on the node transfer device sends process information other than the transfer process on the physical memory area on the main storage device on the node transfer device to the node transfer device. a swap out section to swap out to the virtual memory space on the auxiliary storage device, the Only the information stored in the virtual memory area is characterized by comprising a transfer unit for transferring to the external storage device.

In the present invention, the swap-out unit secures a dummy page in the physical memory area on the main storage device on the node transfer device, so that all the information stored in the physical memory area is stored in the node transfer device. The virtual memory area on the auxiliary storage device is forcibly swapped out.

The present invention, the transfer unit, among the information stored in virtual memory space on the auxiliary storage device on the node transfer device, the virtual memory area process running in said node transfer device is not used All the information other than the above information is transferred to the external storage device.

Further, according to the present invention, in the suspended state in which all processes being executed in the node transfer device other than the node transfer process are suspended, the swap-out unit stores the physical memory area on the main storage device on the node transfer device. After swapping out information other than the information stored in the memory area allocated to the kernel to the virtual memory area on the auxiliary storage device on the node transfer device, the reserved area is allocated to the physical memory area, the information stored in the memory area allocated to the kernel is stored by exporting the reserved area, after all the information stored in the reserved area is swapped out to the virtual memory area, the transfer unit, the virtual that all transfers only the information stored in the memory area in the external storage device And butterflies.

The present invention is also a computer system comprising a node transfer device and a node substitution device that takes over the process of the node transfer device, and includes a storage device external to the node transfer device, and the node The alternative device has the same hardware configuration as the node transfer device, and the node transfer device is suspended on the node transfer device in a suspended state in which all processes being executed in the node transfer device other than the node transfer process are stopped. The control unit swaps out process information other than the transfer process on the physical memory area on the main storage device on the node transfer device to the virtual memory area on the auxiliary storage device on the node transfer device, and forwards only the information stored in the area in said external storage device, the node alternative device, the node In the suspend state where the running process was stopped every at said nodes alternative devices other than the standby process feed, control on the node alternative device, said node transfer device is running in the node transfer device other than the node transfer process All the information stored in the physical memory area on the main storage device on the node transfer device written to the external storage device in the suspended state in which all processes are stopped is stored on the main storage device on the node alternative device. Writing to the physical memory area and all the information stored in the virtual memory area on the auxiliary storage device on the node transfer device to the virtual memory area on the auxiliary storage device on the node alternative device, The node transfer device based on the copy written to the area and the virtual memory area Characterized in that it return the process that was being Oite execution.

Further, the present invention is based on a node transfer device and a copy of a memory area of the node transfer device that has the same hardware configuration as the node transfer device and is written to an external storage device for the node transfer device. the computer as a node transfer device in a computer system composed of a node replacement device take over the process of the node transfer device, in the suspended state of stopping all running processes in the nodes transfer device other than the node transfer process, before Swap-out procedure for swapping out process information other than the transfer process on the physical memory area on the main storage device on the node transfer device to the virtual memory area on the auxiliary storage device on the node transfer device, and the virtual memory area rolling only information stored in the external storage device A transfer procedure which is characterized in that for the execution.

As described above, according to the present invention, a node transfer device and a copy of the memory area of the node transfer device that has the same hardware configuration as the node transfer device and is written to an external storage device for the node transfer device. In the computer system composed of the node alternative device that takes over the process of the node transfer device based on the node transfer device, in the suspended state in which all the processes being executed in the node transfer device other than the node transfer process are stopped, the node transfer device All of the information stored in the physical memory area on the main storage device on the transfer device and the information stored in the virtual memory area on the auxiliary storage device on the node transfer device are written out to an external storage device.
Therefore, there is an effect that the environment of the transfer source node can be directly transferred to the transfer destination node via the external storage device.

Further, according to the present invention, the process of the node transport device has the same hardware configuration as that of the node transport device, and is based on a copy of the memory area of the node transport device written to the external storage device for the node transport device. In the suspended state where the node replacement device that takes over the process has suspended all processes running on the node replacement device other than the node transport standby process, the node transport device has stopped all processes running on the node transport device other than the node transport process In the suspended state, all information stored in the physical memory area on the main storage device on the node transfer device written to the external storage device is written to the physical memory area on the main storage device on the node alternative device, and the node Stored in the virtual memory area on the auxiliary storage device on the transport device Information of Te and write to a virtual memory space on the auxiliary storage device on node alternative apparatus in accordance with the copy exported to the physical memory area and the virtual memory area, returns the process that was being executed at the node transfer device.
Therefore, there is an effect that the environment of the transfer source node can be directly transferred to the transfer destination node via the external storage device.

Hereinafter, the best mode for carrying out the present invention will be described.
First, the basic concept of the present invention will be described.
The present invention employs a method of performing resume on another node having the same configuration by performing hibernation performed in a notebook PC or the like via a network.
Hibernation is one of the node power-off options. When hibernation is instructed, the information stored in the main storage device at that time and various hardware statuses are recorded in the hibernation area of the auxiliary storage device. Then turn off the power. When the power is turned on again, the information recorded in the hibernation area is written again to the main storage device, and the state at the time when hibernation was started last time is restored. As a result, the process that was being executed is restored.
That is, hibernation is based on the premise that suspend and resume are performed at the same node, and a volatile storage area such as a physical memory is recorded on a non-volatile medium such as a built-in hard disk, and resume, especially this is restored.
On the other hand, according to the present invention, the node environment is resumed at a transfer destination node different from the transfer source node. That is, it is necessary to transfer not only the physical memory area of the volatile main storage device but also the memory area not associated with hibernation, such as the virtual memory area (swap area) of the auxiliary storage device, to the transfer destination node. Hereinafter, embodiments of the present invention will be described based on such a concept.

Hereinafter, an embodiment of a computer system including a node transfer device and a node substitution device according to the present invention will be described with reference to the drawings. FIG. 1 is a configuration diagram of a computer system including a node transfer device 1 and a node substitution device 2 according to this embodiment.
As shown in FIG. 1, the computer system of this embodiment is composed of n service providing nodes having the same specifications constituting a PC cluster, and an alternative node (node alternative device 2) having the same specifications as the service providing nodes. The The node transport apparatus 1 serving as a transport source node performs processing as a service providing node during normal operation, and when the service providing node is stopped for maintenance or the like, the service providing node whose operation has been stopped is used as the node transport apparatus 1. .
Then, the node environment of the node transfer device 1 is transferred to the node alternative device 2 and the node transfer device 1 is stopped. In that case, the process or session that was operating in the node transfer apparatus 1 is directly taken over by the node substitution apparatus 2. Furthermore, the node replacement device 2 may be re-transferred to another service providing node in the PC cluster.

That is, the node transport apparatus 1 stores information (including the CPU context) stored in the physical memory area of the main storage device and the virtual memory area (swap area) of the auxiliary storage device as node resources necessary for taking over the processing. The stored information and the file system are all written out to the external storage device 3 different from the node transfer device 1.
Further, the node alternative device 2 has the same hardware configuration as the node transfer device 1 as described above, and based on the copy of the memory area of the node transfer device 1 written to the external storage device 3, Take over the process.

FIG. 2 is a block diagram illustrating the configuration of the node transfer device 1 and the node substitution device 2 according to the present embodiment. As shown in FIG. 2, the node transfer device 1 includes a control unit 11, a main storage device 12, an auxiliary storage device 13, and a network interface unit 14, and the node alternative device 2 includes a control unit 21, The main storage device 22, the auxiliary storage device 23, and the network interface unit 24 are included.
The control unit 11 includes a CPU that controls data input / output in the main storage device 12, the auxiliary storage device 13, and the network interface unit 14, and stops all processes being executed in the node transfer device 1 other than the node transfer process. In the suspended state, as node resources, information (including the CPU context) stored in the physical memory area of the main storage device, information stored in the virtual memory area (swap area) of the auxiliary storage device, and the file system A process of transferring to an external storage device 3 different from the node transfer device 1 is performed.

At this time, all of the information stored in the virtual memory area of the auxiliary storage device 13 is written to the external storage device 3 except for the virtual memory area information not used by the process being executed in the node transfer device 1. Is also possible.
In another embodiment, all information stored in the physical memory area of the main storage device 12 is stored in the auxiliary storage device 13 in the suspended state in which all processes being executed in the node transfer device 1 other than the node transfer process are stopped. It is also conceivable to write the information stored in the virtual memory area to the external storage device 3 after writing to the virtual memory area.
Furthermore, as another embodiment, the kernel (OS) of the information stored in the physical memory area of the main storage device 12 in the suspended state in which all processes being executed in the node transfer device 1 other than the node transfer process are stopped. The information other than the information stored in the memory area allocated to) is written to the virtual memory area of the auxiliary storage device 13, the reserved area is allocated to the physical memory area, and the information stored in the memory area allocated to the kernel is reserved. It is also conceivable that the information stored in the virtual memory area is written to the external storage device 3 after all the information stored in the reserved area is written in the virtual memory area.

The main storage device 12 is configured by a volatile storage device such as a RAM, and stores information necessary for taking over an application process such as a resource ID (process identifier, port number, etc.) and an IP address in a physical memory area.
The auxiliary storage device 13 is a local storage having a virtual memory area (swap area) for complementing the physical memory area of the main storage device 12, and manages the swap area by a paging method or a segment method. In a normal time, the swap area is managed by the OS, and a page being used by a running process and a page not being used are recorded as page management information.
The network interface unit 14 manages data input / output with a network such as a LAN, and is connected to the external storage device 3 configured by a file server or the like.

  Similar to the control unit 11, the control unit 21 includes a main storage device 22, an auxiliary storage device 23, a CPU that controls data input / output in the network interface unit 24, and the like in the node substitution device 2 other than the node transfer standby process. In the suspended state in which all processes being executed are stopped, the node transfer device 1 writes the data to the external storage device 12 in the suspended state in which all processes being executed in the node transfer device 1 other than the node transfer process are stopped. All the information stored in the physical memory area is written to the physical memory area of the main storage device 22 and all the information stored in the virtual memory area of the auxiliary storage device 13 is written to the virtual memory area of the auxiliary storage device 23. Copy and export to physical memory area and virtual memory area Based on, it performs a process of returning the process that was being executed at the node transport device 1.

The main storage device 22 is composed of a volatile storage device such as a RAM, like the main storage device 12, and all information (resource ID (process ID) stored in the physical memory region of the main storage device 12 is stored in the physical memory region. Information necessary for taking over the application process, such as an identifier and a port number) and an IP address.
Similar to the auxiliary storage device 13, the auxiliary storage device 23 is a local storage having a virtual memory area (swap area) for complementing the physical memory area of the main storage device 12.
Similar to the network interface unit 14, the network interface unit 24 manages data input / output with a network such as a LAN, and is connected to the node transfer device 1 via the external storage device 3.

Next, operations of the node transfer device 1 and the node substitution device 2 according to the present embodiment will be described with reference to the drawings. 3 and 4 are a schematic diagram and a sequence diagram of the node transfer processing between the node transfer device 1 and the node alternative device 2 according to the present embodiment.
First, at startup, the node transfer device 1 connects to the external storage device 3, reads a kernel image stored in advance, and starts normal operation.
As shown in FIGS. 3 and 4, during normal operation (step S1 in FIG. 4), when suspend is started (step S2), the node transport apparatus 1 uses the page required by the process being executed in the physical memory area. To swap in and swap out unnecessary pages. Specifically, in response to the suspend instruction, the control unit 11 freezes the process (step S3), and writes the process context of the frozen process in the swap area of the auxiliary storage device 13 (step S4). Next, the control unit 11 forcibly swaps out the information in the physical memory area to the swap area of the auxiliary storage device 13 and connects the information to the external storage device 3 (step S5). (Step S6).

The physical memory area transfer process will be described in detail below.
In the cluster PC, it is rare that the usage amount of the swap area becomes zero. Also, the sum of the physical memory area usage and the swap area usage often exceeds the memory capacity of the physical memory area. Furthermore, since the memory management mechanism of the OS is configured to use the swap area as an extension of the physical memory, the affinity between the swap area and the physical memory is high.
As described above, instead of directly transferring the contents of the physical memory area from the node transfer device 1 to the external storage device 3, all the information stored in the physical memory area is once swapped out to the swap area, and then the entire swap area is Transfer all at once.

The method of writing all physical memory in the swap area once is performed as follows.
That is, the control unit 11 first places all other processes in a special stop state in which processes other than the suspend process do not operate (process freeze).
Next, the control unit 11 secures dummy pages as much as possible on the physical memory.
Then, by the OS memory management mechanism, the control unit 11 swaps out the memory page on the main storage device 12 used by the conventional process to the local swap area on the auxiliary storage device 13 ((( see a)).
However, pages that cannot be swapped out, such as the memory space occupied by the kernel and the buffer for IO, remain in the memory. In this specification, this is called a reserved page.
Therefore, for the reserved page, the control unit 11 secures a swap-out page buffer having the same capacity as the reserved page, and copies each page to the page buffer (see FIG. 5B). Since each page of the page buffer is not a reserved page, all the page buffers are swapped out (see FIG. 5C).
As described above, all the information (memory image) stored in the physical memory area is transferred to the swap area.

Next, swap area transfer processing will be described.
In general, it is rare to use up all the page area of the swap area. This is because the swap area is an auxiliary mechanism that compensates for the shortage of the physical memory area, and there is a problem that the performance is remarkably deteriorated when a system configuration completely dependent on the swap area is adopted. Therefore, the node transport apparatus 1 transfers only the pages in use in the swap area.

Specifically, by incorporating the node transfer function into the OS on the node transfer apparatus 1, the control unit 11 reads the page management data of the swap area managed by the kernel process and uses each page of the swap area. Know the situation.
Then, the control unit 11 selects only the page used by the process being executed based on the usage status of each page in the swap area, and transfers it to the external storage device 3 via the network interface 14. At this time, the control unit 11 maps the transfer record of the selected page and similarly transfers it to the external storage device 3.
As described above, the control unit 11 transfers the suspend image of the swap area to the external storage device 3. Then, after the transfer process is completed, the control unit 11 disconnects from the external storage device 3 and shuts down (steps S7 and S8).

Similar to the node transfer device 1, the node substitution device 2 is connected to the external storage device 3 at the time of activation, and reads a kernel image stored in advance (steps S10 and S11). Next, the node substitution device 2 reads the suspend image of the swap area of the node transfer device 1 stored in the external storage device 3 and disconnects from the external storage device 3.
Then, the node substitution device 2 starts the resume process and restores the process frozen by the node transfer device 1.
Specifically, like the information recorded in the hibernation area, the control unit 21 reads the suspend image into the main storage device 22 and causes the node transfer device 1 to freeze the same as when the hibernation was started last time. Restores the state of the active process. As a result, the process that was being executed is restored.

As described above, according to the node transfer apparatus 1 and the node alternative apparatus 2 of the present embodiment, the node transfer apparatus 1 freezes its environment (physical memory or local file system), and the frozen node environment is networked. To the node alternative device 2 via Then, in the node substitution device 2, the frozen environment is thawed.
Therefore, the effect that the node environment in the node transfer device 1 can be transferred to the node alternative device 2 is obtained. More specifically, according to the node transfer device 1 and the node alternative device 2 of the present embodiment, the environment of the node transfer device 1 is migrated to the node alternative device 2 including the environment of the OS executing the application. . In addition, the process that was running before the transfer is not terminated at all, but is transferred to the transfer destination computer. In addition, before and after the migration, the OS operates in the native environment.
In other words, the function (hibernation) of turning off the power of a node without ending the OS or process in the same node is characterized in that it is not the same node to be restored but a different node.

That is, according to the node transport apparatus 1 and the node alternative apparatus 2 of the present embodiment, in the PC cluster environment, in order to perform hard maintenance on a specific node without stopping the entire service, it is executed on the node. The entire service is continued by transferring the existing process to another standby node. As a result, the internal state of the process that was being executed at the time of migration is retained, so that the problem of limited use of roll-forward applications can be solved, and the effect of realizing a completely transparent migration is obtained. It is done.
In addition, since the address translation at the time of migration does not occur, the effect of minimizing the influence on the external network can be obtained.
In addition, since the CPU power is not consumed by another OS unlike the virtual machine method, the CPU performance can be fully utilized during normal operation, so that it is possible to prevent a reduction in processing capacity. An effect is obtained.
In addition, since it is not necessary for the transfer source node to specify an alternative node as a transfer destination, when the alternative node is managed externally, an effect of simplifying the node transfer management can be obtained.

In the above embodiment, an example in which the suspend image of the swap area of the node transfer device 1 is read from the node alternative device 2 via the external storage device 3 has been described, but the present invention is not limited to this. Alternatively, the suspend image may be directly transferred without going through the external storage device 3. Specifically, as shown in FIG. 6, the node alternative device 2 is set in the transfer standby state (idle state), and the process frozen by the node transfer device 1 is written in the swap area and then transferred. , Read the transferred suspend image and resume.
With this configuration, the node environment can be migrated from the node transfer device 1 to the node alternative device 2 without using the external storage device 3. Therefore, an effect that the hardware cost of the external storage device 3 can be reduced is obtained.

In the above embodiment, the specific example of the node resource is not particularly limited. However, for example, as shown in FIG. 7, when the hierarchical configuration of the node resource is defined, together with the node environment at the application level, the OS level In other words, all node environments can be migrated.
For reference, FIG. 8 shows a node resource hierarchical configuration diagram based on the roll forward method, FIG. 9 a distributed OS type, and FIG. 10 a virtual computer type.
As already mentioned, in the roll-forward method, the application that was executed on the migration source node is newly started on the migration destination node, and the status of the application is made the same as that of the transfer source node using the snapshot. In contrast to the re-execution, in the node transfer device 1 and the node alternative device 2 of the present embodiment, the OS level node environment is also migrated.
Further, in the distributed OS type, the OS operates in cooperation while managing the distributed memory between the nodes, whereas the node transfer device 1 and the node alternative device 2 of the present embodiment do not use the distributed memory.
In the virtual machine type, a virtual machine emulator is used, whereas in the node transfer device 1 and the node alternative device 2 of this embodiment, a virtual computer emulator is not used.

  In the above embodiment, the method for mounting the local file system of the auxiliary storage device 13 on the auxiliary storage device 23 is not particularly limited. For example, the physical memory area of the main memory 12 or the auxiliary storage device 13 described above is used. As in the case of transferring the swap area to the external storage device 3, the local file system of the auxiliary storage device 13 is once written to the external storage device 3 and read by the auxiliary storage device 23 at the time of resume. It is done. As another embodiment, a local file system common to the node transfer device 1 and the node alternative device 2 is written in the external storage device 3 in advance, and similarly, the auxiliary storage device 23 reads the local file system when resuming. It may be.

A series of processes related to the node transfer process and the node substitution process described above are stored in a computer-readable recording medium in the form of a program, and the above process is performed by the computer reading and executing this program. .
That is, each processing means and processing unit in the node transfer device 1 and the node substitution device 2 is a central processing unit such as a CPU that reads the above program into a main storage device such as a ROM or a RAM, and processes and calculates information. This is realized by executing.
Here, the computer-readable recording medium means a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Alternatively, the computer program may be distributed to the computer via a communication line, and the computer that has received the distribution may execute the program.

The block diagram of the computer system comprised from the node transfer apparatus 1 and the node alternative apparatus 2 of this embodiment. The block diagram which shows the structure of the node transfer apparatus 1 of this embodiment, and the node alternative apparatus 2. FIG. Schematic of the node transfer process between the node transfer apparatus 1 and the node alternative apparatus 2 of this embodiment. The sequence diagram of the node transfer process between the node transfer apparatus 1 and the node alternative apparatus 2 of this embodiment. Explanatory drawing which shows a mode that it swaps out from a physical memory area | region to a swap area | region. The sequence diagram which shows a mode that the suspend image of a swap area | region is directly transferred between the node transfer apparatus 1 of this embodiment, and the node alternative apparatus 2. FIG. The hierarchical block diagram of the node transfer apparatus 1 and the node alternative apparatus 2 of this embodiment. FIG. 4 is a hierarchical configuration diagram of roll-forward node resources. FIG. 2 is a hierarchical configuration diagram of a distributed OS type node resource. FIG. 2 is a hierarchical configuration diagram of virtual machine type node resources.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Node transfer apparatus 2 ... Node alternative device 3 ... External storage device 11, 21 ... Control part 12, 22 ... Main storage device 13, 23 ... Auxiliary storage device 14, 24 ... Network interface part

Claims (6)

A node transport device having the same hardware configuration as the node transport device, and a process of the node transport device based on a copy of a memory area of the node transport device written to a storage device external to the node transport device A node transfer device in a computer system comprising a node alternative device that takes over
In a suspended state in which all processes being executed in the node transport device other than the node transport process are stopped,
The control unit on the node transfer device is
A swap-out unit for swapping out process information other than a transfer process on a physical memory area on a main storage device on the node transfer device to a virtual memory area on an auxiliary storage device on the node transfer device;
A transfer unit that transfers only the information stored in the virtual memory area to the external storage device;
A node transfer apparatus comprising:   The swap-out unit is
  All the information stored in the physical memory area is forced to the virtual memory area on the auxiliary storage device on the node transfer device by securing a dummy page in the physical memory region on the main storage device on the node transfer device Swap out,
  The node transfer apparatus according to claim 1. The transfer unit is
Of the information stored in the virtual memory area on the auxiliary storage device on the node transfer device,
Node transfer device according to claim 1 or claim 2, characterized in that all transfer information other than the information of the virtual memory area running process is not used in the node transfer device to said external storage device . In a suspended state in which all processes being executed in the node transport device other than the node transport process are stopped,
The swap-out unit is
Among the information stored in the physical memory area on the main storage device on the node transfer device, the virtual memory area on the auxiliary storage device on the node transfer device other than the information stored in the memory area assigned to the kernel After swapping out
A reserve area is allocated to the physical memory area;
The information stored in the memory area allocated to the kernel is stored by exporting the reserved area, after swapping out all the information stored in the reserved area in the virtual memory area,
The transfer unit is
The node transfer device according to any one of claims 1 to 3, wherein only the information stored in the virtual memory area is transferred to the external storage device. A computer system comprising a node transport device and a node alternative device that takes over the process of the node transport device,
An external storage device for the node transport device;
The node alternative device is
The same hardware configuration as the node transport device,
The node transport device is
In a suspended state in which all processes being executed in the node transport device other than the node transport process are stopped,
The control unit on the node transfer device is
Swapping out process information other than the transfer process on the physical memory area on the main storage device on the node transfer device to the virtual memory area on the auxiliary storage device on the node transfer device ,
Only transfers the information stored in the virtual memory area in said external storage device,
The node alternative device is
In the suspended state in which all processes being executed in the node alternative device other than the node transfer standby process are stopped,
The control unit on the node alternative device is
Stored in the physical memory area on the main storage device on the node transfer device written to the external storage device in the suspended state in which all processes being executed on the node transfer device other than the node transfer process are stopped by the node transfer device And all the information thus written to the physical memory area on the main storage device on the node alternative device,
Writing all the information stored in the virtual memory area on the auxiliary storage device on the node transfer device to the virtual memory area on the auxiliary storage device on the node alternative device;
Based on the copy written to the physical memory area and the virtual memory area, the process being executed in the node transfer device is restored.
A computer system characterized by that. A node transport device having the same hardware configuration as the node transport device, and a process of the node transport device based on a copy of a memory area of the node transport device written to a storage device external to the node transport device In a computer as a node transfer device in a computer system composed of a node alternative device that takes over
In a suspended state in which all processes being executed in the node transport device other than the node transport process are stopped ,
And swap out procedures to swap out to the virtual memory space on the auxiliary storage device of the process information for other transport processes in physical memory area on the main memory on the previous SL nodes transfer device on the node transfer device,
A transfer procedure for transferring only information relevant stored in the virtual memory area in said external storage device,
Node transfer program for causing the execution.

Images