DATA AWARE STORAGE SYSTEM OPERATIONS

• US 20120173831A1
• Filed: 12/29/2010
• Published: 07/05/2012
• Est. Priority Date: 12/29/2010
• Status: Active Grant

• Current Assignee: Teradata US, Inc. (Teradata Corporation)
• Original Assignee: Teradata US, Inc. (Teradata Corporation)
• Inventors: Rubio, Paul Anthony, Carrillo, Criselda Samera, Khobare, Aarati Abhijit, Boggs, Gary Lee, Sanders, Kevin Reed
• Granted Patent: US 8,510,528 B2
• Time in Patent Office: Days
• US Class Current: 711/165
• CPC Class Codes: G06F 3/0605 by facilitating the interaction with a user or administrator
• CPC Class Codes: G06F 3/0649 Lifecycle management
• CPC Class Codes: G06F 3/0685 Hybrid storage combining heterogeneous device types, e.g. hierarchical storage, hybrid arrays

Abstract

Apparatus, systems, and methods may operate to classify storage locations in a storage medium according to at least three response time grades, to classify data to be stored in the storage locations according to at least three access frequency grades, and to migrate the data between the storage locations according to a predicted access frequency assigned to preemptive allocations of some of the storage locations, based on the response time grade and the access frequency grade associated with the data prior to migration. Other apparatus, systems, and methods are disclosed.

Claims

1. An apparatus, comprising:

a storage medium having storage locations; and

a processor to classify the storage locations according to at least three response time grades, to classify data to be stored in the storage locations according to at least three access frequency grades, and to migrate the data between the storage locations according to a predicted access frequency assigned to preemptive allocations of some of the storage locations, based on the response time grade and the access frequency grade associated with the data prior to migration.

2. The apparatus of claim 1, further comprising:

a reservation module to preemptively reserve extents in the storage medium, providing a number of reserved extents based on assigning an access frequency grade that is higher than other ones of the access frequency grades to the reserved extents, wherein the reserved extents are associated with a response time grade that is higher than other ones of the response time grades, prior to storing the data in the reserved extents.

3. The apparatus of claim 1, further comprising.

multiple database processors, including the processor, coupled to corresponding storage nodes that each include some of the data that form a portion of a database.

4. The apparatus of claim 1, further comprising:

a first node to house the storage medium; and

a second node to house the processor comprising an access module processor.

5. The apparatus of claim 1, wherein the storage medium comprises a combination of solid state storage devices and mechanical storage devices.

6. A processor-implemented method to execute on one or more processors that perform the method, comprising:

classifying storage locations in a storage medium according to at least three response time grades;

classifying data to be stored in the storage locations according to at least three access frequency grades; and

migrating the data between the storage locations according to a predicted access frequency assigned to preemptive allocations of some of the storage locations, based on the response time grade and the access frequency grade associated with the data prior to migration.

7. The method of claim 6, wherein the decay modeling comprises:

monitoring a number of allocated extents, de-allocated extents, or a function of the allocated and de-allocated extents, in the storage medium, decayed over time.

8. The method of claim 6, wherein the decay modeling comprises:

monitoring access frequency of the data, decayed over time.

9. The method of claim 6, further comprising:

calculating a benefit lifetime of the data as a time period approximately proportional to an expected time remaining before the data is next moved.

10. The method of claim 6, further comprising:

preemptively reserving a number of reserved extents in the storage medium in an amount approximately proportional to a calculated benefit lifetime of the data multiplied by a difference between a number of allocated extents in the storage medium, decayed over time, and a number of de-allocated extents in the storage medium, decayed over time.

11. The method of claim 6, further comprising:

determining an initial response time grade to be assigned to the data based on a space allocation type associated with the data.

12. The method of claim 6, wherein a number of the response time grades is approximately equal to a number of cylinders in the storage medium.

13. The method of claim 12, further comprising:

migrating the data from a first one of the storage locations to a second one of the storage locations when a benefit of moving the data is greater than a cost of moving the data.

14. The method of claim 13, further comprising:

calculating the benefit of moving the data in proportion to the access frequency grade associated with the data, multiplied by a difference between the response time grade associated with the first one of the storage locations and the response time grade associated with the second one of the storage locations.

15. The method of claim 13, further comprising:

calculating the cost of moving the data in proportion to at least one queue depth associated with the storage medium and a number of input/output operations used to move the data.

16. The method of claim 13, wherein migrating the data further comprises:

selectively moving subsets of the data in an order determined by values of a difference between the benefit minus the cost, wherein the values are calculated for each one of the subsets.

17. The method of claim 6, further comprising:

assigning an access frequency grade that is higher than other ones of the access frequency grades to reserved extents in the storage medium, wherein the reserved extents are associated with a response time grade that is higher than other ones of the response time grades.

18. A processor-implemented method to execute on one or more processors that perform the method, comprising:

classifying storage locations in a storage medium according to at least three response time grades;

classifying data to be stored in the storage locations according to at least three access frequency grades; and

migrating the data between the storage locations according to a predicted access frequency assigned to preemptive allocations of some of the storage locations, based on decay modeling of the response time grade and the access frequency grade associated with the data prior to migration.

19. The method of claim 18, further comprising:

preemptively reserving a number of extents in the storage medium based on a calculated benefit lifetime of the data.

20. The method of claim 18, wherein the decay modeling comprises:

modeling the response time grades and/or the access frequency grades using an exponential decay formula.