Caching system and method providing aggressive prefetch

• US 6,003,114 A
• Filed: 06/17/1997
• Issued: 12/14/1999
• Est. Priority Date: 06/17/1997
• Status: Expired due to Term

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Bachmat, Eitan
• Primary Examiner(s): Chan, Eddie P.
• Assistant Examiner(s): MCLEAN MAYO, KIMBERLY N
• Application Number:
  US08/877,700
• Time in Patent Office: 910 Days
• Field of Search: 711/133, 711/134, 711/135, 711/136, 711/137, 711/117, 711/118, 711/112, 711/113, 711/114, 395/872
• US Class Current: 711/113
• CPC Class Codes: G06F 12/0862 with prefetch
• CPC Class Codes: G06F 12/0866 for peripheral storage systems, e.g. disk cache

Abstract

A mass storage subsystem includes one or more data stores for storing information for access by one or more devices such as host computers, and a cache memory, and connects to a slave mass storage subsystem which provides mirrored storage for the information stored in its (that is, the mass storage subsystem'"'"'s) data stores. When a host computer is to access information, the information is cached in the cache memory by the data stores, and the host computer will retrieve the information from the cache memory. In addition, the host computers can update information through the cache memory. Over a period of time, the mass storage subsystem gathers statistics relating to cache misses for respective chunks, each chunk comprising a relatively large number contiguous storage locations as maintained by the data stores, and will periodically cache information from entire chunks. Generally, if there is a relatively large number of cache misses for a chunk over a period of time, the chunk is being randomly accessed by the host computers, and, by caching the entire chunk, the likelihood of having a cache miss for that chunk will be eliminated. In addition, caching chunks in the slave mass storage subsystem'"'"'s cache memory will increase the utilization of the cache memory of the slave mass storage subsystem.

Claims

1. A mass storage system comprising:

A. a cache memory;

B. a data store configured to store information in chunks each having a plurality of blocks, each block having at least one item of information, and in response to a staging request, perform a staging operation to copy information from one of said blocks into said cache;

C. a host adapter configured to perform access operations in connection with information cached in said cache memory, the host adapter being configured to, during an access operation, initially determine whether the information to be accessed is in the cache memory and, if not, generate a staging request to enable the data store to perform a staging operation to copy the information to be accessed into the cache memory and update cache miss information associated with one of said chunks containing the block; and

D. a chunk cache control configured to periodically process the cache miss information associated with a plurality of said chunks to select said chunks whose information is to be cached, and to enable information from the selected chunks to be copied into the cache memory.

2. A mass storage system as defined in claim 1 in which said host adapter is configured to perform said access operations in response to access requests from a host computer.

3. A mass storage system as defined in claim 1 further comprising a chunk statistics table including at least one entry associated with a respective one of said chunks, said at least one entry including at least one field for receiving cache miss information associated with the respective one of the chunks, the host adapter being further configured to update the cache miss information in the entry associated with the respective chunk.

4. A mass storage system as defined in claim 3 in which:

A. said chunk statistics table includes a plurality of entries each associated with a respective one of said chunks, said entries each including at least one field for receiving cache miss information associated with the respective chunk,

B. the host adapter is further configured to, when it determines that information to be accessed during an access operation is not in the cache memory, update the cache miss information in the entry of the chunk statistics table that is associated with one of the chunks containing the information to be accessed.

5. A mass storage system as defined in claim 3 in which:

A. said chunk statistics table includes a plurality of entries each associated with a respective one of said chunks, said entries each including at least one field for receiving cache miss information associated with the respective chunk,

B. the chunk cache control is further configured to periodically process the cache miss information in said entries of said chunk statistics table to identify the selected chunks whose information is to be cached, and to enable information from the selected chunks to be copied into the cache memory.

6. A mass storage system as defined in claim 3 in which the chunk cache control is further configured to select said chunks whose cache miss information indicates a relatively large number of cache misses over a predetermined time period.

7. A mass storage system as defined in claim 3 in which the chunk cache control is further configured to reset the cache miss information in said entries.

8. A mass storage system is defined in claim 1

A. in which:

i. the cache memory comprises a main cache memory and a chunk cache memory;

ii. the data store is further configured to, in response to a staging request, perform a staging operation to copy information from one of said blocks into said main cache memory;

iii. the host adapter is further configured to perform access operations in connection with information cached in said main cache memory; and

iv. the chunk cache control is further configured to enable information from the selected chunks to be copied into the chunk cache memory;

B. said mass storage subsystem further comprising a cached chunk table identifying the ones of said chunks cached in the chunk cache memory,

C. the host adapter is further configured to, if it determines that the information to be retrieved is not in the main cache memory, use the cached chunk table to determine whether the chunk containing the information to be retrieved is cached in the chunk cache memory, and, if so, generate a transfer request for the chunk cache control; and

D. the chunk cache control is further configured to, in response to a transfer request from the host adapter, transfer the information to be retrieved to the main cache memory.

9. A mass storage system as defined in claim 8 in which said main cache memory, said data store as a master data store and said host adapter form a master mass storage subsystem, said mass storage system further including a slave mass storage subsystem including said chunk cache memory and at least one slave data store, the slave data store storing information in a mirrored storage arrangement mirroring the information stored in said master data store, the chunk cache control being further configured to interconnect said master mass storage subsystem and said slave mass storage subsystem.

10. A method of operating a mass storage system of the type comprising

i. a cache memory;

ii. a data store configured to store information in chunks each having a plurality of blocks, each block having at least one item of information, the data store being further configured to, in response to a staging request, perform a staging operation to copy information from one of said blocks into said cache;

iii. a host adapter configured to perform access operations in connection with information cached in said cache memory, the host adapter being configured to, during an access operation, initially determine whether the information to be accessed is in the cache memory and, if not, generate a staging request to enable the data store to perform a staging operation to copy the information to be accessed into the cache memory and update cache miss information associated with the one of said chunks containing the block;

the method comprising the steps of;

A. updating cache miss information associated with the one of said chunks containing the block when the host adapter determines during an access operation that the information to be accessed is not in the cache memory; and

B. periodically processing the cache miss information associated with a plurality of said chunks to select ones of said chunks whose information is to be cached, and

C. copying information from the selected chunks into the cache memory.

11. A method as defined in claim 10 further comprising the step of providing access requests to the host adapter, the host adapter performs said access operations in response to said access requests.

12. A method as defined in claim 10, the mass storage subsystem further comprising a chunk statistics table including at least one entry associated with a respective one of said chunks, said at least one entry including at least one field for receiving cache miss information associated with the respective one of the chunks, the cache miss information update step including the step of adapter updating the cache miss information in the entry associated with the respective chunk.

13. A method as defined in claim 12, the chunk statistics table including a plurality of entries each associated with a respective one of said chunks, said entries each including at least one field for receiving cache miss information associated with the respective chunk, the cache miss information update step including the step of updating the cache miss information in the entry of the chunk statistics table that is associated with the one of the chunks containing the information to be accessed.

14. A method as defined in claim 12, the chunk statistics table includes a plurality of entries each associated with a respective one of said chunks, said entries each including at least one field for receiving cache miss information associated with the respective chunk, in which

B. during the cache miss information processing step, periodically processing the cache miss information in said entries of said chunk statistics table to identify the selected chunks whose information is to be cached,

C. during the information copying step, copying information from the selected chunks into the cache memory.

15. A method as defined in claim 12 in which, during the cache miss information processing step, the ones of said chunks whose cache miss information indicates a relatively large number of cache misses over a predetermined time period are identified.

16. A method as defined in claim 12 further comprising the step of resetting the cache miss information in said entries after said cache miss information processing step.

17. A method as defined in claim 10, the method being for use in connection with the mass storage system

i. in which:

a) the cache memory comprises a main cache memory and a chunk cache memory;

b) the data store is further configured to, in response to a staging request, perform a staging operation to copy information from one of said blocks into said main cache memory;

c) the host adapter is further configured to perform access operations in connection with information cached in said main cache memory; and

d) the chunk cache control is further configured to enable information from the selected chunks to be copied into the chunk cache memory;

ii. said mass storage subsystem further comprising a cached chunk table identifying the ones of said chunks cached in the chunk cache memory,

the method further comprising the steps of;

A. in response to a determination that the information to be retrieved is not in the main cache memory, using the cached chunk table to determine whether the chunk containing the information to be retrieved is cached in the chunk cache memory; and

B. if so, transfer the information to be retrieved from the chunk cache memory to the main cache memory.

18. A method as defined in claim 17, the method being for use in connection with the mass storage system in which said main cache memory, said data store as a master data store and said host adapter form a master mass storage subsystem, said mass storage system further including a slave mass storage subsystem including said chunk cache memory and at least one slave data store, the slave data store storing information in a mirrored storage arrangement mirroring the information stored in said master data store, in which, during the information copying step, copying the information from the slave data store to the chunk cache memory.