Algorithm methodologies for efficient compaction of overprovisioned memory systems

• US 10,489,288 B2
• Filed: 03/27/2017
• Issued: 11/26/2019
• Est. Priority Date: 01/25/2017
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Malladi, Krishna T., Zheng, Hongzhong
• Primary Examiner(s): Mercado, Ramon A.
• Application Number:
  US15/470,261
• Publication Number:
  US 20180210659A1
• Time in Patent Office: 974 Days
• CPC Class Codes: G06F 12/023 Free address space management
• CPC Class Codes: G06F 2212/1044 Space efficiency improvement

Abstract

A method of dynamically selecting deduplication granularity in a memory system to decrease deduplication granularity and to increase hash-table efficiency, the method including selecting one or more deduplication granularities at an application level of an application using the memory system, the one or more deduplication granularities being selected according to features of the memory system, and assigning a memory region corresponding to each of the one or more selected deduplication granularities, where the method may use a memory manager to share memory translation table and hash table, and may be employed by a system that enables using higher capacity pre-allocated counter fields for frequently utilized lines.

Claims

1. A method of dynamically selecting deduplication granularity in a memory system comprising a cache to decrease deduplication granularity and to increase hash-table efficiency, the method comprising:

selecting one or more deduplication granularities at an application level of an application using the memory system based at least in part on historical data associated with the memory system, and dynamically assigning the one or more deduplication granularities for a cache line comprising a unit of data transfer, the one or more deduplication granularities being selected according to features of the memory system; and

assigning a memory region corresponding to the one or more selected deduplication granularities.

2. The method of claim 1, wherein selecting the one or more deduplication granularities comprises using a software driver corresponding to the application using the memory system.

3. The method of claim 2, further comprising rebooting the memory system to implement parameters resulting from the selection of the one or more deduplication granularities by the software driver.

4. The method of claim 1, wherein selecting the one or more deduplication granularities comprises partitioning a memory space into memory regions, the memory regions corresponding to the one or more deduplication granularities.

5. The method of claim 4, wherein partitioning the memory space into memory regions is performed by a kernel driver of the memory system.

6. The method of claim 1, wherein a deduplication system-on-a-chip (SoC) on a deduplication memory module of the memory system is used for observing the historical data associated with the memory system.

7. The method of claim 1, wherein the historical data associated with the memory system comprise an application pattern history, a deduplication ratio, or projected table overheads.

8. A method of dynamically selecting deduplication granularity in a memory system comprising a cache to decrease deduplication granularity and to increase hash-table efficiency, the method comprising:

selecting one or more deduplication granularities at an application level of an application using the memory system, the one or more deduplication granularities being selected according to features of the memory system; and

assigning a memory region corresponding to the one or more selected deduplication granularities,

wherein selecting the one or more deduplication granularities comprises;

partitioning a memory space into two memory regions, the memory regions corresponding to two deduplication granularities as the one or more deduplication granularities;

performing both of the two deduplication granularities on an incoming cache line comprising a unit of data transfer;

storing a table in the two memory regions for the deduplication granularities; and

discarding a table entry of one of the tables depending on learned history when a combined capacity of both of the tables of the two memory regions either increases or reaches a threshold.

9. The method of claim 8, wherein a deduplication system-on-a-chip (SoC) on a deduplication memory module of the memory system is used for performing both of the two deduplication granularities on the incoming cache line, and for storing the table in the two memory regions.

10. The method of claim 8, wherein selecting the one or more deduplication granularities comprises using a software driver corresponding to the application using the memory system.

11. The method of claim 10, further comprising rebooting the memory system to implement parameters resulting from the selection of the one or more deduplication granularities by the software driver.

12. The method of claim 8, wherein selecting the one or more deduplication granularities comprises partitioning a memory space into memory regions, each of the memory regions corresponding to the one or more deduplication granularities.

13. The method of claim 12, wherein partitioning the memory space into memory regions is performed by a kernel driver of the memory system.

14. A system architecture for deploying a deduplication memory module, and for being controlled by a driver or a system-on-a-chip (SoC) to:

select one or more deduplication granularities at an application level of an application based at least in part on historical data associated with a memory system comprising a cache, and dynamically assigning the one or more deduplication granularities for a cache line comprising a unit of data transfer, the one or more deduplication granularities being selected according to features of the memory system; and

assign a memory region corresponding to each of the one or more selected deduplication granularities.