Parallel computer, migration program and migration method

Patent metadata

Field	Value
Publication number	US-2016357442-A1
Application number	US-201615143654-A
Country	US
Kind code	A1
Filing date	May 2, 2016
Priority date	Jun 3, 2015
Publication date	Dec 8, 2016
Grant date	—

Abstract

A parallel computer includes a first node and a second node, each including a memory having a plurality of memory areas and a cache memory, and a processing unit that acquires a first group of index levels of the cache memory, the first group of index levels corresponding with addresses of first plurality of memory areas storing data accessed by a job in the first node, when continuing an execution of the job by migrating the job carried out on the first node to the second node, judges whether or not the second node has second plurality of memory areas that are a usable state corresponding to a second group of index levels that has same as or relative position relation with the first group of index levels, and relocates the data to the second plurality of memory areas when the second node has the second plurality of memory areas.

First claim

What is claimed is:

1 . A parallel computer comprising:
a first node and a second node, each including:
a memory having a plurality of memory areas; and
a cache memory, and
a processing unit that
acquires a first group of index levels of the cache memory in the first node, the first group of index levels corresponding with addresses of first plurality of memory areas that store data accessed by a job in the first node, when continuing an execution of the job by migrating the job that carried out on the first node to the second node,
judges whether or not the second node has second plurality of memory areas that are a usable state corresponding to a second group of index levels that has same as or relative position relation with the first group of index levels that is acquired, and
relocates the data to the second plurality of memory areas when the second node has the second plurality of memory areas.

2 . The parallel computer according to claim 1 , wherein the second group of index levels having the relative position relation comprises a group of index levels which is shifted the first group of index levels uniformly.

3 . The parallel computer according to claim 1 , wherein the processing unit carries out the acquisition and the judgment for a different node from the second node when the second node has not the second plurality of memory areas.

4 . The parallel computer according to claim 1 , wherein the processing unit, when there are a plurality of target candidate nodes of migration destination of the job executed by the first node, calculates a judgment value indicating an outbreak degree of a slashing after the migration based on state about use of a plurality memory areas in the candidate nodes corresponding to the second group of index levels for each of the plurality of target candidate nodes, and relocates the data to the second plurality of memory areas in the second node that has smaller judgment value among the plurality of target candidate nodes.

5 . The parallel computer according to claim 4 , wherein the processing unit calculates a first judgment value when the state about use of the plurality of memory areas in the candidate node indicates a non-use state, and a second judgment value that is larger than the first judgment value when the state about use of the plurality of memory areas in the candidate node includes memory area that is a first state indicating a use state that is not accessed for a predetermined period.

6 . The parallel computer according to claim 5 , wherein the processing unit calculates a third judgment value which is smaller than the second judgment value and is larger than the first judgment value when the state about use of the plurality of memory areas in the candidate node includes memory area that is a second state indicating a temporally use.

7 . The parallel computer according to claim 6 , wherein the processing unit calculates a fourth judgment value which is larger than the second judgment value when the state about use of the plurality of memory areas in the candidate node includes memory area that is a third state indicating a use state by an operating system.

8 . The parallel computer according to claim 4 , the processing unit acquires a third group of index levels of a cache memory corresponding with addresses of third plurality of memory areas that store data accessed by other job in a third node,
judges whether or not the first node has fourth plurality of memory areas that are the usable state corresponding to a fourth group of index levels that has same as or relative position relation with the third group of index levels after relocating the data, and relocates the data accessed by other job to the third plurality of memory areas when the first node has the third plurality of memory areas.

9 . A non-transitory computer readable storage medium storing therein a program that causes a computer to execute a process, the process comprising:
acquiring a first group of index levels of the cache memory in a first node, the first group of index levels corresponding with addresses of first plurality of memory areas that store data accessed by a job in the first node, when continuing an execution of the job by migrating the job that carried out on the first node to the second node; judging whether or not the second node has second plurality of memory areas that are a usable state corresponding to a second group of index levels that has same as or relative position relation with the first group of index levels that is acquired; and relocating the data to the second plurality of memory areas when the second node has the second plurality of memory areas.

10 . The storage medium according to claim 9 , wherein the second group of index levels having the relative position relation comprises a group of index levels which is shifted the first group of index levels uniformly.

11 . The storage medium according to claim 9 , wherein the acquiring and the judging comprises acquiring and judging for a different node from the second node when the second node has not the second plurality of memory areas.

12 . The storage medium according to claim 9 , wherein the process further comprises:
calculating, when there are a plurality of target candidate nodes of migration destination of the job executed by the first node, a judgment value indicating an outbreak degree of a slashing after the migration based on state about use of a plurality memory areas in the candidate nodes corresponding to the second group of index levels for each of the plurality of target candidate nodes; and relocating the data to the second plurality of memory areas in the second node that has smaller judgment value among the plurality of target candidate nodes.

13 . The storage medium according to claim 12 , wherein the calculating comprises:
first calculating a first judgment value when the state about use of the plurality of memory areas in the candidate node indicates a non-use state; and second calculating a second judgment value that is larger than the first judgment value when the state about use of the plurality of memory areas in the candidate node includes memory area that is a first state indicating a use state that is not accessed for a predetermined period.

14 . The storage medium according to claim 13 , wherein the calculating further comprises third calculating a third judgment value which is smaller than the second judgment value and is larger than the first judgment value when the state about use of the plurality of memory areas in the candidate node includes memory area that is a second state indicating a temporally use.

15 . A migration method comprising:
acquiring, by a processor, a first group of index levels of the cache memory in a first node, the first group of index levels corresponding with addresses of first plurality of memory areas that store data accessed by a job in the first node, when continuing an execution of the job by migrating the job that carried out on the first node to a second node; judging, by a processor, whether or not the second node has second plurality of memory areas that are a usable state corresponding to a second group of index levels that has same as or relative position relation with the first group of index levels that is acquired; and relocating, by a processor, the data to the second plurality of memory areas when the second node has the second plurality of memory areas.

16 . The migration method according to claim 15 , wherein the second group of index levels having the relative position relation co

Assignees

• Fujitsu Ltd

Inventors

• Ninomiya Atsushi

Classifications

• G06F2212/1024

  Latency reduction · CPC title

• G06F3/0647

  Migration mechanisms · CPC title

• G06F2212/62

  Details of cache specific to multiprocessor cache arrangements · CPC title

• G06F2212/2542

  Non-uniform memory access [NUMA] architecture · CPC title

• G06F9/485

  Task life-cycle, e.g. stopping, restarting, resuming execution (G06F9/4881 takes precedence) · CPC title