System and Methodology Providing Multiple Heterogeneous Buffer Caches

• US 20080140937A1
• Filed: 12/12/2006
• Published: 06/12/2008
• Est. Priority Date: 12/12/2006
• Status: Active Grant

• Current Assignee: SAP America Incorporated (SAP SE)
• Original Assignee: Sybase Incorporated (SAP SE)
• Inventors: Pasupuleti, KantiKiran K., Nalawade, Vaibhav A., Bhatt, Vadiraja P.
• Granted Patent: US 7,831,772 B2
• Time in Patent Office: Days
• US Class Current: 711/119
• CPC Class Codes: G06F 12/0871 Allocation or management of cache space
• CPC Class Codes: G06F 12/0882 Page mode
• CPC Class Codes: G06F 12/122 of the least frequently used [LFU] type, e.g. with individual count value
• CPC Class Codes: G06F 2212/283 Plural cache memories

Abstract

A method for temporarily storing data objects in memory of a distributed system comprising a plurality of servers sharing access to data comprises steps of: reserving memory at each of the plurality of servers as a default data cache for storing data objects; in response to user input, allocating memory of at least one of the plurality of servers as a named cache reserved for storing a specified type of data object; in response to an operation at a particular server requesting a data object, determining whether the requested data object is of the specified type corresponding to the named cache at the particular server; if the data object is determined to be of the specified type corresponding to the named cache, storing the requested data object in the named cache at the particular server; and otherwise, using the default data cache for storing the requested data object.

Claims

1. In a distributed system comprising a plurality of servers sharing access to data, a method for temporarily storing data objects in memory of said distributed system, the method comprising:

reserving a portion of memory at each of said plurality of servers as a default data cache for temporarily storing data objects;

in response to user input, allocating a portion of memory of at least one of said plurality of servers as a named cache reserved for temporarily storing a specified type of data object;

in response to an operation at a particular server requesting a data object, determining whether the requested data object is of the specified type corresponding to the named cache at the particular server;

if the data object is determined to be of the specified type corresponding to the named cache at the particular server, temporarily storing the requested data object in the named cache at the particular server; and

otherwise, using the default data cache at the particular server for temporarily storing the requested data object.

2. The method of claim 1, wherein said distributed system comprises a distributed database system wherein said plurality of servers share access to a single database.

3. The method of claim 2, wherein the operation at the particular server comprises a database query.

4. The method of claim 1, wherein said allocating step includes binding a data object of said specified type to the named cache.

5. The method of claim 4, wherein said user input includes executing a stored procedure for binding data objects to the named cache.

6. The method of claim 4, wherein said determining step includes consulting named cache binding information maintained at the particular server.

7. The method of claim 4, wherein the named cache is allocated in memory at a plurality of servers and binding the data object includes binding the data object to the named cache at each of said plurality of servers.

8. The method of claim 7, further comprising:

in response to failover of a first server, temporarily storing data objects bound to the named cache at the first server in a corresponding named cache at a second server.

9. The method of claim 1, wherein said determining step includes determining whether the named cache exists at the particular server.

10. The method of claim 1, further comprising:

partitioning the named cache into a plurality of separate memory pools;

wherein at least some of said plurality of separate memory pools are of different sizes;

prior to temporarily storing a requested data object in the named cache, estimating size of the requested data object; and

selecting a particular memory pool for temporarily storing the requested data object based on its estimated size.

11. The method of claim 1, wherein said allocating step includes allocating a plurality of different named caches and binding objects of specified types to each of said plurality of named caches.

12. The method of claim 11, wherein said allocating a plurality of different named caches includes allocating more than one different named caches on a particular server, so that data objects of different types are allocated to different named caches on the particular server.

13. The method of claim 1, wherein said temporarily storing step includes temporarily storing at least one data page including the requested data object in the named cache.

14. The method of claim 1, wherein said temporarily storing step includes:

maintaining a queue of data objects temporarily stored in the named cache at a particular server based on how recently each data object is used, wherein least-recently used data objects are displaced from the named cache by more-recently used data objects.

15. The method of claim 1, wherein said temporarily storing step includes reading the requested data object into the named cache from disk storage.

16. The method of claim 15, further comprising:

before reading the requested data object into the named cache from disk storage, determining whether the requested data object is already stored in the named cache; and

if the requested data object is already stored in the named cache, accessing the data object in the named cache for performing the operation, so as to avoid reading the requested data object from disk storage.

17. A computer-readable medium having processor-executable instructions for performing the method of claim 1.

18. The method of claim 1, further comprising:

downloading a set of processor-executable instructions for performing the method of claim 1.

19. In a distributed database system comprising a plurality of servers sharing access to data, a system for storing data objects in memory, the system comprising:

a named cache comprising an allocated portion of memory of at least one of said plurality of servers for storing specified data objects;

a module for binding data objects to the named cache based on user input;

a module for determining, in response to an operation at a particular server requesting a data object, whether the requested data object has a binding to the named cache at the particular server;

a module for storing the requested data object in the named cache at the particular server if the data object is determined to have a binding to the named cache; and

a default data cache comprising a reserved portion of memory at each of said plurality of servers for temporarily storing data objects without a binding to the named cache.

20. The system of claim 19, wherein said distributed database system comprises plurality of database servers sharing access to a single database on disk storage.

21. The system of claim 19, wherein the operation at the particular server comprises a Structured Query Language (SQL) command submitted by a client to said particular server.

22. The system of claim 19, wherein the module for binding provides for binding data objects of user specified types to the named cache.

23. The system of claim 19, wherein said user input includes executing a stored procedure for binding specified data objects to the named cache.

24. The system of claim 19, wherein the module for determining consults named cache binding information maintained at the particular server.

25. The system of claim 19, wherein the named cache is allocated in memory at more than one server and said module for binding the data object binds the data object to the named cache at each of said more than one server.

26. The system of claim 25, wherein said module for storing at a first server operates in response to failover of a second server by storing data objects bound to the named cache at the second server in a corresponding named cache at the first server.

27. The system of claim 19, wherein the module for determining determines whether the named cache exists at the particular server.

28. The system of claim 19, wherein the named cache is partitioned into a plurality of separate memory pools wherein at least some of said plurality of separate memory pools are of different sizes and the module for storing selects a memory pool for storing the requested data object based on estimated size of the requested data object.

29. The system of claim 28, wherein the named cache partitioning is user configurable such that the user can specify sizes of said separate memory pools.

30. The system of claim 28, wherein the named cache is allocated on more than one server and the named cache is partitioned differently on different servers.

31. The system of claim 19, wherein said module for storing step stores at least one data page including the requested data object in the named cache.

32. The system of claim 19, wherein said module for storing maintains a queue of data objects stored in the named cache at a particular server based on how recently each data object is used, wherein least-recently used data objects are displaced from the named cache by more-recently used data objects.

33. The system of claim 19, wherein said module for storing reads the requested data object into the named cache from disk storage.

34. The system of claim 33, further comprising:

a module for determining whether the requested data object is already stored in the named cache before reading the requested data object into the named cache from disk storage, so as to avoid reading the requested data object from disk storage if it is already in the named cache.

35. A system for managing storage of data objects in a distributed database system, the system comprising:

one or more storage devices on which data objects of a database are persistently stored;

a plurality of database servers sharing access to the database, wherein each of said plurality of database servers includes memory for temporarily storing data objects;

a portion of the memory of at least one of said plurality of database servers allocated as a named buffer cache for specified data objects, so as to provide more efficient access to said specified data objects in response to database operations;

a module for binding specified data objects to the named buffer cache based on user input, so that the specified data objects will be stored in the named buffer cache during database operations; and

a module for storing a particular data object requested by a database operation at a given database server in the named buffer cache if said particular database object is bound to the name cache available at the given database server.

36. The system of claim 35, wherein said user input includes executing a stored procedure for binding specified data objects to the named cache.

37. The system of claim 35, wherein the named buffer cache is allocated in memory at more than one database server and said module for binding binds the specified data objects to the named cache at each of said more than one database server.

38. The system of claim 37, wherein said module for storing at a first server operates in response to failover of a second server by storing data objects bound to the named buffer cache at the second server in a corresponding named buffer cache at the first server.

39. The system of claim 35, wherein the named cache is partitioned into a plurality of separate memory pools wherein at least some of said plurality of separate memory pools are of different sizes and the module for storing selects a memory pool for storing the particular data object based on estimates of its size.

40. The system of claim 39, wherein the named buffer cache partitioning is user configurable such that the user can specify sizes of said separate memory pools.

41. The system of claim 35, wherein said module for storing retains data objects in the named buffer cache at a particular server based on how recently each data object is used such that least-recently used data objects are displaced from the named buffer cache by more-recently used data objects.

42. The system of claim 35, wherein said module for storing reads the requested data object into the named buffer cache from disk storage.

43. The system of claim 42, further comprising:

a module for determining whether the requested data object is already stored in the named buffer cache before reading the requested data object into the named buffer cache from disk storage, so as to avoid reading the requested data object from disk storage if it is already in the named buffer cache.