Infrastructure for spilling pages to a persistent store

• US 8,352,517 B2
• Filed: 03/02/2009
• Issued: 01/08/2013
• Est. Priority Date: 03/02/2009
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Park, Hoyong, Jain, Namit, Srinivasan, Anand, Mishra, Shailendra
• Primary Examiner(s): AL HASHEMI, SANA A
• Application Number:
  US12/395,871
• Publication Number:
  US 20100223305A1
• Time in Patent Office: 1,408 Days
• Field of Search: 707/708, 707/822, 707/825, 707/828, 707/758, 707/809
• US Class Current: 707/809
• CPC Class Codes: G06F 16/24542 Plan optimisation
• CPC Class Codes: G06F 16/24547 Optimisations to support specific applications; Extensibility of optimisers

Abstract

Techniques for managing memory usage in a processing system are provided. This may be achieved by receiving a data stream including multiple tuples and determining a query plan that was generated for a continuous query applied to the multiple tuples in the data stream. The query plan may include one or more operators. Before scheduling an operator in the query plan, it is determined when an eviction is to be performed based a level of free memory of the processing system. An eviction candidate is determined and a page associated with the eviction candidate is evicted from the memory to a persistent storage.

Claims

1. A method for managing memory usage in a processing system, the method comprising:

receiving a data stream comprising a plurality of tuples;

determining a query plan generated for a continuous query applied to the plurality of tuples in the data stream, the query plan including one or more operators;

determining a level of free memory associated with the processing system;

prior to scheduling an operator of the one or more operators, determining whether the operator of the one or more operators can be scheduled based at least in part on the level of free memory of the processing system;

in response to determining that the operator of the one or more operators cannot be scheduled, determining an eviction candidate; and

evicting, from the memory to a persistent storage, a page associated with the eviction candidate.

2. The method of claim 1, wherein the evicted page is comprised of one or more tuples to be processed by the eviction candidate.

3. The method of claim 1, further comprising:

storing the evicted page to the persistent storage.

4. The method of claim 1, wherein the processing system is multi-threaded.

5. The method of claim 1, further comprising:

sending an instruction to stop execution of the one or more operators.

6. The method of claim 1, further comprising:

comparing a threshold with the level of free memory; and

determining that eviction is required where the level of free memory falls below the threshold.

7. The method of claim 1, further comprising:

determining whether operators have been scheduled for the data stream;

comparing the level of free memory with a range comprising a first threshold specified in an evict policy and a second threshold specified in the evict policy; and

determining that eviction is required based on the comparison using the range.

8. The method of claim 1, wherein the eviction candidate is determined based on a hierarchical position in the query plan.

9. The method of claim 8, wherein operators at a low hierarchical position in the query plan are identified as eviction candidates.

10. The method of claim 1, further comprising:

identifying a page identifier for the page associated with the eviction candidate, wherein eviction is performed using the page identifier.

11. The method of claim 1, further comprising:

executing the operator;

requesting a first tuple of a one or more tuples to be processed by the operator;

determining whether the first tuple is stored in the memory using a page identifier; and

retrieving the first tuple using the page identifier.

12. The method of claim 10, wherein the page identifier corresponds to a page pointer that refers to a page in the memory which includes the first tuple.

13. The method of claim 10, wherein the first tuple is retrieved from the persistent storage using the page identifier, where the page pointer refers to a null value.

14. The method of claim 1, wherein eviction is selectively performed when a range window size of a query exceeds a size threshold.

15. A computer program product embedded in a computer readable storage medium and including executable instructions for managing memory usage of the processing system, the computer program product comprising:

computer code for receiving a data stream comprising a plurality of tuples;

computer code for determining a query plan generated for a continuous query applied to the plurality of tuples in the data stream, the query plan including one or more operators;

computer code for determining a level of free memory associated with the processing system;

computer code for prior to scheduling an operator of the one or more operators, determining whether the operator of the one or more operators can be scheduled based at least in part on the level of free memory of the processing system;

computer code for determining an eviction candidate in response to determining that the operator of the one or more operators cannot be scheduled; and

computer code for evicting, from the memory to a persistent storage, a page associated with the eviction candidate.

16. The computer program product of claim 15, further comprising:

computer code for sending an instruction to stop execution of the one or more operators.

17. The computer program product of claim 15, further comprising:

computer code for comparing a threshold with the level of free memory; and

computer code for determining that eviction is required where the level of free memory falls below the threshold.

18. A system for managing memory usage of the processing system, the system including a processor and a memory device including instructions that, when executed by the processor, cause the processor to:

receive a data stream comprising a plurality of tuples;

determine a query plan generated for a continuous query applied to the plurality of tuples in the data stream, the query plan including one or more operators;

determine a level of free memory associated with the processing system;

prior to scheduling an operator of the one or more operators, determine whether the operator of the one or more operators can be scheduled based at least in part on the level of free memory of the processing system;

determine an eviction candidate in response to determining that the operator of the one or more operators cannot be scheduled; and

evict, from the memory to a persistent storage, a page associated with the eviction candidate.

19. The system according to claim 18, wherein the instructions, when executed by the processor, further cause the processor to:

send an instruction to stop execution of the one or more operators.

20. The system according to claim 18, wherein the instructions, when executed by the processor, further cause the processor to:

compare a threshold with the level of free memory; and

determine that eviction is required where the level of free memory falls below the threshold.