Automatic caching of scan and random access data in computing systems

• US 10,229,161 B2
• Filed: 09/17/2014
• Issued: 03/12/2019
• Est. Priority Date: 09/20/2013
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Kakarla, Sarat B., Shi, Jia, Aya, Selcuk, Umamageswaran, Kothanda, Loaiza, Juan R.
• Primary Examiner(s): Waldron, Scott A.
• Assistant Examiner(s): Lodhi, Andalib F
• Application Number:
  US14/489,221
• Publication Number:
  US 20150088805A1
• Time in Patent Office: 1,637 Days
• CPC Class Codes: G06F 16/217 Database tuning G06F16/2282 takes precedence; database performance monitoring G06F11/3409
• CPC Class Codes: G06F 16/24552 Database cache management
• CPC Class Codes: G06F 16/283 Multi-dimensional databases or data warehouses, e.g. MOLAP or ROLAP

Abstract

Approaches, techniques, and mechanisms are disclosed for improved caching in database systems that deal with multiple data access patterns, such as in database systems that interface with both OLTP and Data Warehouse clients. A cache is deployed between a database server and a storage system that stores data units. Some of the data units accessed by the database server are buffered within the cache. The data units may be associated with data access patterns, such as a random data access pattern or a scan data access pattern, in accordance with which the database server is or appears to be accessing the data units. A processor selects when to cache data units accessed by the database server, based at least on the associated data access patterns. Recent access counts may also be stored for the data units, and may further be utilized to select when to cache data units.

Claims

1. A method comprising:

responsive to requests from a database server to access data units stored in one or more storage systems, determining, by a caching system comprising a cache, that at least certain data units accessed by the requests are not found in the cache;

determining whether particular data units of the certain data units are associated with a scan data access pattern, wherein the scan data access pattern is indicative that the requests from the database server are for a scan operation that involves accessing substantially more data units than is typical of an Online Transaction Processing (OLTP) operation;

in response to determining that the particular data units of the certain data units are associated with the scan data access pattern;

applying one or more first caching policies to the particular data units to determine whether or not to cache the particular data units;

in response to determining, based on the one or more first caching polices, to cache the particular data units, storing at least one of the particular data units in the cache responsive to the requests; and

in response to determining, based on the one or more first caching policies, to not cache the particular data units, not storing at least one of the particular data units in the cache responsive to the requests;

in response to determining that the particular data units of the certain data units are not associated with the scan data access pattern;

applying one or more second caching policies to the particular data units to determine whether or not to cache the particular data units, wherein the one or more second caching policies are different from the one or more first caching policies; and

in response to determining, based on the one or more second caching policies, to cache the particular data units, storing at least one of the particular data units in the cache responsive to the requests;

wherein the method is performed by one or more computing devices.

2. The method of claim 1, further comprising:

tracking recent access counts for the data units stored in the one or more storage systems, the recent access counts being based in part on how frequently the data units have been accessed over a recent period of time, the one or more first caching policies being based at least partially on the recent access counts.

3. The method of claim 1, further comprising:

tracking recent access counts for the data units stored in the one or more storage systems, the recent access counts being based in part on how frequently the data units have been accessed over a recent period of time, the one or more first caching policies being based at least partially on the recent access counts; and

adjusting the recent access counts to emphasize data units associated with a random access pattern more heavily than data units associated with the scan data access pattern.

4. The method of claim 1, further comprising:

tracking recent access counts for the data units stored in the one or more storage systems, the recent access counts being based in part on how frequently the data units have been accessed over a recent period of time.

5. The method of claim 1, further comprising:

tracking recent access counts for the data units stored in the one or more storage systems, the recent access counts being based in part on how frequently the data units have been accessed over a recent period of time;

wherein applying the one or more first caching policies comprises determining whether to cache a given data unit associated with the scan access data pattern by comparing the recent access count of the given data unit to one or more recent access counts for a set of cached data units already stored in the cache, and comparing a last access time of the given data unit to one or more last access times for the set of cached data units already stored in the cache.

6. The method of claim 5, further comprising adjusting the recent access counts to emphasize more recent accesses of the data units more heavily than less recent accesses.

7. The method of claim 1, wherein the scan data access pattern pertains to data accessed for data warehousing operations, and a random data access pattern pertains to data accessed for OLTP operations.

8. The method of claim 1, further comprising applying the one or more second caching polices to always cache any data unit associated with a random data access pattern.

9. The method of claim 1, wherein applying the one or more first caching policies comprises determining in which cache lines, of a plurality of cache lines within the cache, to store a given data unit, based at least in part on which particular data access pattern is associated with the given data unit.

10. The method of claim 1,

wherein applying the one or more first caching policies to a given data unit associated with the scan data access pattern includes determining to cache the given data unit if a number of cache lines in the cache are determined to be eligible to store the given data unit, the number being of sufficient size to store the given data unit;

wherein determining a cache line to be eligible comprises;

when existing data already stored in the cache line is associated with the scan data access pattern, comparing a recent access count of the existing data to a recent access count of the given data unit; and

when the existing data already stored in the cache line is associated with a random data access pattern, comparing a recent access count of the existing data to a weighted recent access count of the given data unit.

11. The method of claim 1, wherein applying the one or more first caching policies to a given data unit associated with the scan data access pattern further includes determining to cache the given data unit only if doing so will not result in the cache being comprised of more than a threshold percentage of data associated with the scan data access pattern.

12. The method of claim 1, further comprising associating a given data unit of the certain data units with a given data access pattern based on one or more of:

an access pattern indicator in metadata associated with a most recent request in which the given data unit is involved, a characteristic of the most recent request, or a characteristic of the given data unit.

13. One or more non-transitory computer-readable media storing instructions that, when executed by the one or more computing devices, cause performance of:

responsive to requests from a database server to access data units stored in one or more storage systems, determining, by a caching system comprising a cache, that at least certain data units accessed by the requests are not found in the cache;

determining whether particular data units of the certain data units are associated with a scan data access pattern, wherein the scan data access pattern is indicative that the requests from the database server are for a scan operation that involves accessing substantially more data units than is typical of an Online Transaction Processing (OLTP) operation;

in response to determining that the particular data units of the certain data units are associated with the scan data access pattern;

applying one or more first caching policies to the particular data units to determine whether or not to cache the particular data units; and

in response to determining, based on the one or more first caching policies, to cache the particular data units, storing at least one of the particular data units in the cache responsive to the requests; and

in response to determining, based on the one or more first caching policies, to not cache the particular data units, not storing at least one of the particular data units in the cache responsive to the requests;

in response to determining that the particular data units of the certain data units are not associated with the scan data access pattern;

applying one or more second caching policies to the particular data units to determine whether or not to cache the particular data units, wherein the one or more second caching policies are different from the one or more first caching policies; and

in response to determining, based on the one or more second caching policies, to cache the particular data units, storing at least one of the particular data units in the cache responsive to the requests.

14. The one or more non-transitory computer-readable media of claim 13, wherein the instructions, when executed by the one or more computing devices, further cause performance of:

tracking recent access counts for the data units stored in the one or more storage systems, the recent access counts being based in part on how frequently the data units have been accessed over a recent period of time, the one or more first caching policies being based at least partially on the recent access counts.

15. The one or more non-transitory computer-readable media of claim 13, wherein the instructions, when executed by the one or more computing devices, further cause performance of:

tracking recent access counts for the data units stored in the one or more storage systems, the recent access counts being based in part on how frequently the data units have been accessed over a recent period of time, the one or more first caching policies being based at least partially on the recent access counts; and

adjusting the recent access counts to emphasize data units associated with a random access pattern more heavily than data units associated with the scan data access pattern.

16. The one or more non-transitory computer-readable media of claim 13, wherein the instructions, when executed by the one or more computing devices, further cause performance of:

tracking recent access counts for the data units stored in the one or more storage systems, the recent access counts being based in part on how frequently the data units have been accessed over a recent period of time.

17. The one or more non-transitory computer-readable media of claim 13, wherein the instructions, when executed by the one or more computing devices, further cause performance of:

tracking recent access counts for the data units stored in the one or more storage systems, the recent access counts being based in part on how frequently the data units have been accessed over a recent period of time;

wherein applying the one or more first caching policies comprises;

determining whether to cache a given data unit associated with the scan access data pattern by comparing the recent access count of the given data unit to one or more recent access counts for a set of cached data units already stored in the cache, and comparing a last access time of the given data unit to one or more last access times for the set of cached data units already stored in the cache.

18. The one or more non-transitory computer-readable media of claim 17, wherein the instructions further cause adjusting the recent access counts to emphasize more recent accesses of the data units more heavily than less recent accesses.

19. The one or more non-transitory computer-readable media of claim 13, wherein the scan data access pattern pertains to data accessed for data warehousing operations, and a random data access pattern pertains to data accessed for OLTP operations.

20. The one or more non-transitory computer-readable media of claim 13, wherein the one or more second caching policies includes a third policy to always cache any data unit associated with a random data access pattern.

21. The one or more non-transitory computer-readable media of claim 13, wherein applying the one or more first caching policies comprises determining in which cache lines, of a plurality of cache lines within the cache, to store a given data unit, based at least in part on which particular data access pattern is associated with the given data unit.

22. The one or more non-transitory computer-readable media of claim 13,

wherein applying the one or more first caching policies to a given data unit associated with the scan data access pattern includes determining to cache the given data unit if a number of cache lines in the cache are determined to be eligible to store the given data unit, the number being of sufficient size to store the given data unit;

wherein determining a cache line to be eligible comprises;

when existing data already stored in the cache line is associated with the scan data access pattern, comparing a recent access count of the existing data to a recent access count of the given data unit; and

when the existing data already stored in the cache line is associated with a random data access pattern, comparing a recent access count of the existing data to a weighted recent access count of the given data unit.

23. The one or more non-transitory computer-readable media of claim 13, wherein applying the one or more first caching policies to a given data unit associated with the scan data access pattern further includes determining to cache the given data unit only if doing so will not result in the cache being comprised of more than a threshold percentage of data associated with the scan data access pattern.

24. The one or more non-transitory computer-readable media of claim 13, wherein the instructions, when executed by the one or more computing devices, further cause performance of:

associating a given data unit of the certain data units with a given data access pattern of the data access patterns based on one or more of;

an access pattern indicator in metadata associated with a most recent request in which the given data unit is involved, a characteristic of the most recent request, or a characteristic of the given data unit.

25. A computer system comprising:

one or more processors;

one or more interfaces communicably coupled to one or more database servers and one or more storage systems;

one or more storage media configured to store a cache comprising cache lines in which are cached data units, from the one or more storage systems, that were accessed by the database server;

one or more storage media configured to store cache metadata, the cache metadata including at least recent access counts reflecting how frequently the data units have been accessed over a recent period of time;

wherein the one or more processors are configured to;

responsive to requests from the one or more database servers to access data units stored in the one or more storage systems, determine, by a caching system comprising the cache, that at least certain data units accessed by the requests are not found in the cache;

determine whether particular data units of the certain data units are associated with a scan data access pattern, wherein the scan data access pattern is indicative that the requests from the database server are for a scan operation that involves accessing substantially more data units than is typical of an Online Transaction Processing (OLTP) operation;

in response to determining that the particular data units of the certain data units are associated with the scan data access pattern;

apply one or more first caching policies to the particular data units to determine whether or not to cache the particular data units;

in response to determining, based on the one or more first caching policies, to cache the particular data units, store at least one of the particular data units in the cache; and

in response to determining, based on the one or more first caching policies, to not cache the particular data units, not store at least one of the particular data units in the cache responsive to the requests;

in response to determining that the particular data units of the certain data units are not associated with the scan data access pattern;

apply one or more second caching policies to the particular data units to determine whether or not to cache the particular data units, wherein the one or more second caching policies are different from the one or more first caching policies; and

in response to determining, based on the one or more second caching policies, to cache the particular data units, store at least one of the particular data units in the cache responsive to the requests.

26. The computer system of claim 25, wherein the one or more processors are further configured to select where to store cacheable data units based at least on the associated data access patterns and the recent access counts.