STORAGE SYSTEM, NODE APPARATUS, CACHE CONTROL METHOD AND PROGRAM

• US 20160266844A1
• Filed: 09/30/2014
• Published: 09/15/2016
• Est. Priority Date: 09/30/2013
• Status: Active Grant

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: OGAWA, Shugo
• Granted Patent: US 10,452,300 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 12/0808 with cache invalidating means G06F12/0815 takes precedence
• CPC Class Codes: G06F 12/0813 with a network or matrix configuration
• CPC Class Codes: G06F 12/0815 Cache consistency protocols
• CPC Class Codes: G06F 12/0891 using clearing, invalidating or resetting means
• CPC Class Codes: G06F 12/0893 Caches characterised by their organisation or structure
• CPC Class Codes: G06F 2212/1016 Performance improvement
• CPC Class Codes: G06F 2212/1024 Latency reduction
• CPC Class Codes: G06F 2212/152 Virtualized environment, e.g. logically partitioned system
• CPC Class Codes: G06F 2212/60 Details of cache memory
• CPC Class Codes: G06F 3/0613 in relation to throughput
• CPC Class Codes: G06F 3/0647 Migration mechanisms
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]
• CPC Class Codes: G06F 3/0685 Hybrid storage combining heterogeneous device types, e.g. hierarchical storage, hybrid arrays
• CPC Class Codes: G06F 9/4856 resumption being on a different machine, e.g. task migration, virtual machine migration G06F9/5088 takes precedence

Abstract

Each node includes a cache to store data of the storage shared by the plurality nodes. Time information when a process accessing to data migrates from one node to another node is recorded. The one node, after migration of the process to the other node, selectively invalidates data held in the cache of the one node with a time of last access thereto by the process on the one node being older than a time of migration of the process from the one node to the other node.

Claims

1. A storage system comprising:

a plurality of nodes; and

a storage shared by the plurality of nodes,

wherein each of the nodes includes

a cache that holds temporarily data stored in the storage,

wherein time information when a process accessing data migrates from one node to other node is recorded, and

the one node, after migration of the process from the one node to the other node, selectively invalidates, out of data held in the cache of the one node, data accessed at least by the process on the one node with a time of last access thereto by the process being older than a time of the migration of the process from the one node to the other node.

2. The storage system according to claim 1, wherein the one node is configured to executes the invalidation at a preset timing following a time when the process migrating from the one node to the other node, returns again to the one node.

3. The storage system according to claim 1, wherein the one node is configured to execute the selective invalidation of data, with an access request to the data held in the cache of the one node with a time of last access thereto by the process on the one node being older than the time of the migration of the process from the one node to the other node, as a trigger, the access request to the data being made by the process that returns to the one node after migrating to the other node.

4. The storage system according to claim 1, wherein the one node is configured to exclude data held in the cache of the one node and accessed, after the migration of the process from the one node to the other node, by a process other than the process migrating to the other node, from a subject of invalidation in the cache of the one node.

5. The storage system according to claim 1, wherein the one node is configured to invalidate as well, data held in the cache of the one node with a time of last access thereto by a process other than the process migrating to the other node being older than the time of the migration of the process migrating to the other node.

6. The storage system according to claim 1, wherein access to data in the storage is executed from a single node at one time, for a write in a cache in the single node, cache invalidation in another node being necessitated.

7. The storage system according to claim 1, wherein when a node executing a first process differs from a node executing a second process sharing data with the first process, the first process or the second process is migrated between the nodes so that the first process and the second process is executed in the same node.

8. The storage system according to claim 1, wherein the node includes

a notification unit that informs a management unit about a start of execution of a process accessing data, wherein

the management unit includes:

an execution node change detection unit that detects a change of the node executing a process accessing data;

an execution node recording unit that records information concerning a node that executes the process;

a time recording unit that records a time when the process accessing the data on the node that executes the process is executed last time;

a process arrangement unit that locates a process so that a plurality of processes accessing data shared by the processes are sequentially executed in the same node; and

an invalidation notification unit that notifies the node about the invalidation of data held in the cache,

the invalidation notification unit informing the node about a time of the change of the node executing the process.

9. The storage system according to claim 1, wherein the node includes a virtualization mechanism, the process being executed on a virtual machine on the virtualization mechanism, and wherein

a change of the node that executes the process is implemented by inter-node migration of the virtual machine.

10. A node apparatus including a cache that temporarily stores data of a storage shared by a plurality of node apparatuses, the node apparatus comprising:

an access time recording unit that records a time of an access each time the access is made to data held in the cache; and

a cache invalidation unit that compares a time of last access to the data held in the cache, recorded in the access time recording unit with a invalidation reference time received and that selectively invalidates data with a time of last access anterior to the invalidation reference time.

11. The node apparatus according to claim 10, wherein the invalidation reference time is a time when a process accessing data migrates from the node apparatus as a source of migration to another node apparatus.

12. The node apparatus according to claim 11, wherein the cache invalidation unit selectively invalidates data held in the cache with a time of last access thereto being anterior to the invalidation reference time, at a timing subsequent to a time when the process migrating from the one node apparatus to the other node apparatus, again returns to the one node apparatus.

13. The node apparatus according to claim 12, wherein the cache invalidation unit selectively invalidates the data, with an access request made by the process returning again to the one node apparatus after migrating to the other node, to data held in the cache thereof with a time of last access thereto by the process on the node apparatus being older than a time of the migration of the process from the one node apparatus to the other node apparatus, as a trigger.

14. A management apparatus connected to a plurality of nodes, each of which includes a cache that temporarily holds data of a storage shared by the nodes;

the management apparatus receiving from the node a notification about the start of execution of the process of data accessing;

the management apparatus comprising;

an execution node change detection unit that detects a change of the node executing the process of data accessing;

an execution node recording unit that records information concerning the node that executes the process;

a time recording unit that records a time when the process accessing the data is executed last time in the node that executes the process;

a process arrangement unit that sets a location of the process so that a plurality of access processes sharing the data are sequentially executed in the same node; and

an invalidation notification unit that informs the node about invalidation of data held in the cache of the node,

the invalidation notification unit informing the node about the time of the change of the node that executes the access process.

15. A method for controlling a cache comprising:

recording time information when a process accessing data migrates from one node of a plurality of nodes to other node, each of the plurality of nodes including a cache in which data stored in a storage shared by the plurality of nodes is held temporarily; and

after migration of the process from the one node to the other node, the one node selectively invalidating, out of data held in the cache of the one node, data accessed at least by the process on the one node with a time of last access thereto by the process being older than a time of the migration of the process from the one node to the other node, at a preset timing or with a preset trigger.

16. The cache controlling method according to claim 15, comprising

the one node executing the invalidation at a timing following a time when the process migrating from the one node to the other node again returns to the one node.

17. The cache controlling method according to claim 15, comprising

the one node executing the selective invalidation of data, with an access request to the data held in the cache with a time of last access thereto by the process on the one node being older than a time of migration of the process from the one node to the other node, as a trigger, the access request to the data being made by the process that returns again to the one node after migrating to the other node.

18. The cache controlling method according to claim 15, comprising

the one node excluding data held in the cache of the one node and accessed, after the migration of the process from the one node to the other node, by a process other than the process migrating from the one node to the other node from a subject of the invalidation in the cache of the one node.

19. The cache controlling method according to claim 15, comprising

the one node invalidating as well, data held in the cache of the one node with a time of last access thereto by a process other than the process that migrates to the other node being older than a time of the migration of the process to the other node.

20. The cache controlling method according to claim 15, comprising

executing access to data held in the storage by a single node at least at one time, for a write in a cache in the single node, cache invalidation being necessitated in another node.

21. The cache controlling method according to claim 15, comprising

when a node executing a first process differs from a node executing a second process sharing data with the first process,

migrating the first process or the second process between the nodes so that the first process and the second process are executed in the same node.

22. The cache controlling method according to of claim 15, wherein the node includes a virtualization mechanism, the method comprising:

executing the access process on a virtual machine on the virtualization mechanism; and

implementing migration of the process by inter-node migration of the virtual machine.

23. A non-transitory computer-readable medium storing therein a program that causes a computer constituting a node apparatus including a cache that temporarily holds data stored in a storage shared by a plurality of node apparatuses to execute the processing comprising:

an access time recording processing that records an access time each time an access is made to data held in the cache; and

a cache invalidation processing that compares a last access time to data held in the cache, recorded by the access time recording processing with an invalidation reference time received and that invalidates data held in the cache with the last access time being anterior to the invalidation reference time, at a preset timing or by a trigger before the data held in the cache is accessed.

24. The non-transitory computer-readable medium according to claim 23, wherein the invalidation reference time is a time when a process accessing data migrates from the node apparatus as a source node of migration to other node apparatus.

25. The non-transitory computer-readable medium according to claim 24, wherein the cache invalidation processing is executed on the node apparatus at a timing subsequent to a time when the process migrating from the node apparatus to the other node apparatus again returns to the node apparatus.

26. The non-transitory computer-readable medium according to claim 24, wherein the cache invalidation processing executes the selective invalidation of data, with an access request to the data held in the cache with a time of last access thereto by the process on the node apparatus being older than a time of the migration of the process from the node apparatus to the other node apparatus, as a trigger, the access request to the data being made by the process that again returns to the node apparatus after migrating to the other node apparatus.