Managing compression in data storage systems

• US 8,478,731 B1
• Filed: 03/31/2010
• Issued: 07/02/2013
• Est. Priority Date: 03/31/2010
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Throop, Dean D., Duprey, Dennis T., Owen, Karl M.
• Primary Examiner(s): TIMBLIN, ROBERT M
• Application Number:
  US12/751,455
• Time in Patent Office: 1,189 Days
• Field of Search: 707/693, 707/999.101
• US Class Current: 707/693
• CPC Class Codes: G06F 16/1744 using compression, e.g. sparse files

Abstract

A method is used in managing compression in data storage systems. A data temperature is determined for a set of data in a block-based data storage system. Based on the data temperature, data compression is applied to the set of data.

Claims

1. A method for use in managing compression in data storage systems, the method comprising:

determining a first data temperature for sub logical unit extents in a block-based data storage system, wherein the block-based data storage system comprises at least one logical unit including sub logical unit extents;

based on the first data temperature, determining a first compression level for the sub logical unit extents using cut off temperatures for separating different compression levels, the compression levels comprising an uncompressed level, a fast compression level and a slow compression level, wherein the fast compression level has a greater performance profile than the slow compression level and the uncompressed level has a greater performance profile than the fast compression level;

based on the first compression level, applying data compression to the sub logical unit extents;

determining a second data temperature for the sub logical unit extents;

generating a compression update candidate list, wherein the compression update candidate list includes an entry for each sub logical unit extent, further wherein the compression update candidate list comprises the second data temperature and a new second compression level for each sub logical unit extent in the list;

dividing the compression update candidate list into separate logical unit candidate lists for each logical unit;

adding the hottest sub logical unit extent in each of the logical unit candidate lists to an execution list; and

updating the compression levels of each sub logical unit extent in the execution list in order of expected performance gain.

2. The method of claim 1, wherein the data temperature is derived from a level of data access activity from data utilization statistics.

3. The method of claim 1, wherein no compression is used for often accessed data and compression is used for infrequently accessed data.

4. The method of claim 1, wherein data access statistics on sub LUN extents are kept and used to rank data from most frequently accessed to less frequently accessed.

5. The method of claim 1, wherein a ranking of data from most frequently accessed to less frequently accessed is consulted and a compression process is used only with the less frequently accessed data.

6. The method of claim 1, wherein accesses to frequently accessed data are able to reference uncompressed data.

7. The method of claim 1, wherein the data of the block-based data storage system is subdivided into slices and a data temperature statistic is kept for each slice.

8. A system for use in managing compression in data storage systems, the system comprising:

a processor comprising a program logic for carrying out the steps of;

first logic determining a first data temperature for sub logical unit extents in a block-based data storage system, wherein the block-based data storage system comprises at least one logical unit including sub logical unit extents;

based on the first data temperature, second logic determining a first compression level for the sub logical unit extents using cut off temperatures for separating different compression levels, the compression levels comprising an uncompressed level, a fast compression level and a slow compression level, wherein the fast compression level has a greater performance profile than the slow compression level and the uncompressed level has a greater performance profile than the fast compression level;

based on the first compression level, third logic applying data compression to the sub logical unit extents;

fourth logic determining a second data temperature for the sub logical unit extents;

fifth logic generating a compression update candidate list, wherein the compression update candidate list includes an entry for each sub logical unit extent, further wherein the compression update candidate list comprises the second data temperature and a new second compression level for each sub logical unit extent in the list;

sixth logic dividing the compression update candidate list into separate logical unit candidate lists for each logical unit;

seventh logic adding the hottest sub logical unit extent in each of the logical unit candidate lists to an execution list; and

eighth logic updating the compression levels of each sub logical unit extent in the execution list in order of expected performance gain.

9. The system of claim 8, wherein the data temperature is derived from a level of data access activity from data utilization statistics.

10. The system of claim 8, wherein no compression is used for often accessed data and compression is used for infrequently accessed data.

11. The system of claim 8, wherein data access statistics on sub LUN extents are kept and used to rank data from most frequently accessed to less frequently accessed.

12. The system of claim 8, wherein a ranking of data from most frequently accessed to less frequently accessed is consulted and a compression process is used only with the less frequently accessed data.

13. The system of claim 8, wherein accesses to frequently accessed data are able to reference uncompressed data.

14. The system of claim 8, wherein the data of the block-based data storage system is subdivided into slices and a data temperature statistic is kept for each slice.