Methods and mechanisms for proactive memory management

• US 7,185,155 B2
• Filed: 06/09/2005
• Issued: 02/27/2007
• Est. Priority Date: 10/04/2002
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Fortin, Michael R., Ergan, Cenk, Iyigun, Mehmet, Sechrest, Stuart
• Primary Examiner(s): Thai; Tuan V.
• Application Number:
  US11/150,728
• Publication Number:
  US 20050228964A1
• Time in Patent Office: 628 Days
• Field of Search: 711/100, 711/133, 711/134, 711/135, 711/136, 711/137, 711/154, 711/159, 711/160, 711/151, 711/170
• US Class Current: 711/158
• CPC Class Codes: G06F 12/121 using replacement algorithms
• CPC Class Codes: G06F 12/122 of the least frequently used [LFU] type, e.g. with individual count value

Abstract

A proactive, resilient and self-tuning memory management system and method that result in actual and perceived performance improvements in memory management, by loading and maintaining data that is likely to be needed into memory, before the data is actually needed. The system includes mechanisms directed towards historical memory usage monitoring, memory usage analysis, refreshing memory with highly-valued (e.g., highly utilized) pages, I/O pre-fetching efficiency, and aggressive disk management. Based on the memory usage information, pages are prioritized with relative values, and mechanisms work to pre-fetch and/or maintain the more valuable pages in memory. Pages are pre-fetched and maintained in a prioritized standby page set that includes a number of subsets, by which more valuable pages remain in memory over less valuable pages. Valuable data that is paged out may be automatically brought back, in a resilient manner. Benefits include significantly reducing or even eliminating disk I/O due to memory page faults.

Claims

1. A tangible computer-readable storage medium having computer-executable instructions stored thereon, which, when executed, perform steps for fetching data from a storage location into memory based on a ranking of an estimated utility for the data, comprising:

observing a need for page data to be stored in memory;

determining values for the page data based on the observed need for the page data, each value representative of the observed need for that page data and independent of any physical memory location; and

loading at least some of the page data into the memory in a prioritized manner based on the determined values for the page data.

2. The computer-readable storage medium of claim 1, wherein loading at least some of the page data into the memory comprises selecting page data based on the determined values.

3. The computer-readable storage medium of claim 1, wherein loading at least some of the page data into the memory comprises loading the entire memory.

4. The computer-readable storage medium of claim 1, wherein loading at least some of the page data into the memory comprises loading the page data into memory that is free.

5. The computer-readable storage medium of claim 1, further comprising, identifying low-value page data in memory, and wherein loading at least some of the page data into the memory comprises replacing at least some of the low value page data.

6. The computer-readable storage medium of claim 1, wherein observing the need for page data comprises tracking actual usage of page data, and/or training to determine likely needed page data that may be stored in memory.

7. The computer-readable storage medium of claim 1, wherein observing the need for data that correspond to pages comprises training to determine likely needed pages, wherein the training includes performing at least one action of a set of actions, the actions including, simulating page use, analyzing code, and running programs on a training computer system and using the values on another computer system.

8. The computer-readable storage medium of claim 1, further comprising, building a persistent repository of information including the determined values, and storing the persistent repository in a file.

9. The computer-readable storage medium of claim 1, wherein determining the values comprises maintaining rich use information, the rich use information comprising at least one of pattern, frequency, grouping and context information for the page data to be stored in memory.

10. A computer-implemented method for efficiently transferring units of data corresponding approximately to a size of a memory page, or page data, from a storage location to a memory location, and vice versa, based on a determined priority for the page data, comprising:

observing a need for one or more sets of page data, to be stored in memory;

determining values for the page data, each value based on the observed need for the corresponding set of page data and independent of any physical memory location;

associating the determined values with the page data; and

determining what of the one or more sets of page data to load from a storage location into the memory based on the determined values associated with the one or more sets of page data.

11. The computer-implemented method of claim 10, wherein determining the values further comprises maintaining rich use information, comprising at least one of pattern, frequency, grouping and context information for the page data in the one or more sets.

12. The computer-implemented method of claim 11, further comprising, reprioritizing the page data based on the rich use information.

13. The computer-implemented method of claim 11, wherein maintaining the rich use information comprises keeping working set entries containing the rich use information, and/or allocating physical page descriptors and entries for maintaining the rich use information for page data that leave memory.

14. The computer-implemented method of claim 10, further comprising maintaining information for page datas that have been but are no longer in the memory, and for page data that have not been in the memory.

15. The computer-implemented method of claim 10, wherein observing the need for sets of page data comprises accessing maintained statistics from which the values can be determined, the statistics comprising at least one of pattern, frequency, grouping and context information.

16. The computer-implemented method of claim 15, wherein accessing the maintained statistics comprises accessing at least one statistical item of a set of page data, the set containing, per data entries, entries for groups of page data, entries for files, and state data.

17. The computer-implemented method of claim 10, further comprising, using the determined values to prioritize page data in memory into subsets.

18. The computer-implemented method of claim 10, further comprising using the determined values to protect more valuable page data from being overwritten with less valuable page data.

19. The computer-implemented method of claim 10, wherein observing the need for one or more sets of page data comprises tracking actual usage of pages data, and/or training to determine likely needed page data.

20. The computer-implemented method of claim 10, further comprising, identifying low-value page data stored in memory, and wherein determining what sets of page data to load into the memory based on the values associated with the page data comprises replacing at least some of the low-value pages data in memory with sets of page data associated with higher determined values than the low-value pages data.

21. In a computerized system comprising a memory that is subdivided into one or more page locations, or memory pages, a method of transferring data, or page data, from one or more storage locations to one or more page locations based on one or more value determinations of the page data, such that high priority data are accessible from the one or more memory pages before they are needed, comprising:

identifying data stored that are stored in one or more pages of memory;

observing, via a tracing mechanism, one or more memory-related activities for the data, and logging results of the observation into an information database, wherein the information database includes information observed for present and prior memory activities on the identified data;

prioritizing the identified data into a prioritization list based on the observations logged in the information database, wherein the prioritization list subdivides the data into a plurality of page data sets each having a unique prioritization value; and

replacing lower value page data in the one or more memory pages with higher value page data.

22. The method as recited in claim 21, further comprising:

observing additional one or more memory-related activities for the high value page data in the one or more memory pages;

rebalancing the prioritization list so that at least some of the high value page data in the one or more memory pages is ascribed lower prioritization value; and

replacing at least one of the at least some of the high value page data with higher value page data in the corresponding one or more memory pages.