Power management in a storage array

• US 7,814,351 B2
• Filed: 06/28/2007
• Issued: 10/12/2010
• Est. Priority Date: 06/28/2007
• Status: Expired due to Fees

• Current Assignee: Seagate Technology LLC (Seagate Technology Holdings Plc)
• Original Assignee: Seagate Technology LLC (Seagate Technology Holdings Plc)
• Inventors: Lubbers, Clark Edward, DeCenzo, David Peter
• Primary Examiner(s): ELAMIN, ABDELMONIEM I
• Application Number:
  US11/770,672
• Publication Number:
  US 20090006877A1
• Time in Patent Office: 1,202 Days
• Field of Search: 713/300, 713/310, 713320-324, 713/330, 713/340, 711/6, 711/114
• US Class Current: 713/300
• CPC Class Codes: G06F 11/34 Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
• CPC Class Codes: G06F 3/0625 Power saving in storage systems
• CPC Class Codes: G06F 3/0634 by changing the state or mode of one or more devices
• CPC Class Codes: G06F 3/0653 Monitoring storage devices or systems
• CPC Class Codes: G06F 3/0659 Command handling arrangements, e.g. command buffers, queues, command scheduling
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]
• CPC Class Codes: G06F 3/0689 Disk arrays, e.g. RAID, JBOD
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A data storage system and associated method is provided within an enclosure supporting a self-contained plurality of discrete data storage devices configured for connecting with a network device via a network. The data storage system includes a redundant array of independent drives (RAID) container services module in the enclosure that allocates and manages a storage space of the data storage devices for storing primary and redundant data, and a policy engine in the enclosure that continuously and qualitatively characterizes the network load to the data storage system and manages a power distribution to each of the data storage devices based on a predicted utilization that differentiates between access commands for primary data and access commands for redundant data.

Claims

1. A data storage system comprising:

an enclosure supporting a self-contained plurality of discrete data storage devices configured for connecting with a network device via a network;

a redundant array of independent drives (RAID) container services module in the enclosure that allocates and manages a storage space of the data storage devices for storing primary and redundant data; and

a policy engine in the enclosure that continuously and qualitatively characterizes the network load to the data storage system and manages a power distribution to each of the data storage devices based on a predicted utilization that differentiates between access commands for primary data and access commands for redundant data.

2. The data storage system of claim 1 wherein the policy engine maintains a statistical record of an observed number of access commands for the primary data as differentiated from access commands for the redundant data to each of the data storage devices during a predetermined interval.

3. The data storage system of claim 1 wherein the policy engine predicts future intervals during which a data storage device is statistically likely to be nonutilized independently of the network, and when a predicted interval is greater than a predetermined threshold governs a power distribution module to reduce a power mode of the data storage device.

4. The data storage subsystem of claim 1 wherein the policy engine characterizes the load in terms of what access commands are read commands in real time.

5. The data storage subsystem of claim 1 wherein the RAID container services module allocates the data storage devices into a primary data device and a mirror data device paired mirroring arrangement for storing data with a RAID-1 fault tolerance, and wherein the policy engine predicts intervals during which the primary data device is statistically likely to receive only read commands, and reduces a power mode of the mirror data device when a predicted read-only command interval is greater than a predetermined threshold.

6. The data storage subsystem of claim 1 wherein the policy engine identifies low frequency access command activity that interrupts a reduced power mode state, and transfers the data blocks associated with the low frequency activity to consolidate identified low frequency command activity to a single data storage device.

7. The data storage subsystem of claim 6 wherein the policy engine transfers the data blocks associated with the low frequency access command activity to cache.

8. The data storage subsystem of claim 1 wherein the policy engine comprises a finite state machine.

9. A method for managing power in a pool of data storage devices, comprising:

placing the pool of data storage devices in a common enclosure with a policy engine in the enclosure capable of characterizing network access commands;

monitoring time of day information for access commands to each data storage device differentiated between access commands for RAID primary data as opposed to access commands for RAID redundant data;

calculating a statistical record of the actual utilization of each data storage device in the pool from the monitoring step findings;

predicting future utilizations of each data storage device from results of the calculating step.

10. The method of claim 9 further comprising allocating the data storage devices in a RAID independently of the storage area network.

11. The method of claim 9 wherein the monitoring step comprises maintaining a statistical record of an observed number of the differentiated access commands to each of the data storage devices during a predetermined interval.

12. The method of claim 9 wherein the predicting step comprises predicting, independently of the network, future intervals during which a data storage device is statistically expected to be nonutilized with respect to RAID redundant data, and when a predicted nonutilization interval is greater than a predetermined threshold reducing a power mode of the data storage device.

13. The method of claim 9 wherein the monitoring step comprises ascertaining what portion of the access commands are read commands and the calculating step comprises calculating a read-only mode utilization of the data storage devices.

14. The method of claim 10 wherein the allocating step comprises providing a paired mirroring arrangement of primary data storage devices and respective mirror data storage devices with RAID-1 fault tolerance, wherein the predicting step comprises projecting a future interval during which the primary data storage device is statistically expected to be only in a read-only utilization mode, further comprising reducing a power mode of the respective mirror storage device when the projected future interval is greater than a predetermined threshold.

15. The method of claim 9 wherein the monitoring step identifies low frequency access command activity interrupting a reduced power mode state, and in response transfers the data blocks associated with the low frequency access command activity to consolidate them in a single data storage device.

16. The method of claim 15 wherein the transferring step comprises transferring the data blocks associated with the low frequency access command activity to a selected one of another data storage device in the pool and another data storage device outside the pool.

17. The method of claim 16 wherein the transferring step comprises transferring the data blocks associated with the low frequency access command activity to cache.

18. The method of claim 9 wherein the monitoring step is characterized by time of day information in terms of at least one of a number of read commands, a number of write commands, a read bandwidth, and a write bandwidth.