Correlated failure zones for data storage

• US 9,110,797 B1
• Filed: 06/27/2012
• Issued: 08/18/2015
• Est. Priority Date: 06/27/2012
• Status: Expired due to Fees

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Lazier, Colin L.
• Primary Examiner(s): DUNCAN, MARC M
• Application Number:
  US13/535,281
• Time in Patent Office: 1,147 Days
• Field of Search: 714/6.11, 714/6.12, 714/6.13
• US Class Current: 1/1
• CPC Class Codes: G06F 11/0727 in a storage system, e.g. in a DASD or network based storage system drivers for digital recording or reproducing units G06F3/06; circuits for error detection or correction within digital recording or reproducing units G11B20/18; for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS], H04L67/1097
• CPC Class Codes: G06F 11/1048 using arrangements adapted for a specific error detection or correction feature
• CPC Class Codes: G11B 20/1883 Methods for assignment of alternate areas for defective areas

Abstract

Techniques for optimizing data storage are disclosed herein. In particular, methods and systems for implementing redundancy encoding schemes with data storage systems are described. The redundancy encoding schemes may be scheduled according to system and data characteristics. The schemes may span multiple tiers or layers of a storage system. The schemes may be generated, for example, in accordance with a transaction rate requirement, a data durability requirement or in the context of the age of the stored data. The schemes may be designed to rectify entropy-related effects upon data storage. The schemes may include one or more erasure codes or erasure coding schemes. Additionally, methods and systems for improving and/or accounting for failure correlation of various components of the storage system, including that of storage devices such as hard disk drives, are described.

Claims

1. A computer-implemented method for optimizing data storage, comprising:

selecting, by one or more computer systems, an electromechanical data storage device that comprises at least;

at least one platter, each platter further comprising two physical platter sides that each comprise a data storage medium that is capable of storing data across a plurality of physical data locations of the data storage medium;

a common spindle threaded through a center of the at least one platter as an axis to enable rotation of the at least one platter;

a first arm having a first head, the first arm associated with a first physical side of the platter, the first head capable of floating above the first physical platter side of the at least one platter while rotating about the common spindle;

a second arm having a second head, the second arm associated with a second physical side of the platter, the second physical side distinct from the first physical side, the second head capable of floating above the second physical platter side of the at least one platter while rotating about the common spindle; and

a controller that, using at least firmware, activates at least one of the first arm with the first head or the second arm with the second head to respond to data write requests by storing data associated with the data write requests across the physical data locations of the data storage medium on the two platter sides; and

determining, by the one or more computer systems, one or more modes of failure of the at least one platter;

designating, by at least the firmware, a plurality of zones based at least in part on the one or more modes of failure, each zone comprising a subset of the physical data locations of each of the physical sides of the platters, the subset having a reduced failure correlation with the remaining physical data locations of the data storage medium;

responding, using at least the firmware, to the data write requests by activating the at least one of the first arm with the first head or the second arm with the second head to store the associated data within a specified zone among the plurality of zones, the data write requests including the specified zone to which the associated data should be written; and

at a time after detecting a malfunction of at least one of the first arm with the first head or the second arm with the second head causing one or more physical data locations within the specified zone to be defective, remapping, by at least the firmware, the defective physical data locations to one or more different physical data locations within the specified correlated fault zone, such that a logical addressing scheme referencing the defective physical data locations is remapped to instead reference the one or more different physical data locations within the specified zone.

2. The computer-implemented method of claim 1, wherein each zone corresponds to a single platter surface of one of the platters.

3. The computer-implemented method of claim 1, further comprising populating the at least a subset of the remapped different physical data locations are populated with data when the defective physical data locations are remapped to the different physical data locations.

4. The computer-implemented method of claim 3, wherein the associated data has been transformed by an erasure coding scheme that causes the increased data durability.

5. The computer-implemented method of claim 1, wherein determining one or more failure modes comprises detecting one or more entropic effects related to data stored on the physical data locations on the platters.

6. The computer-implemented method of claim 5, wherein the one or more entropic effects includes performance degradation due to at least one of data fragmentation resulting from repeated write and re-write cycles, firmware degradation, data unavailability or loss due to malfunction, external environmental factors, or physical destruction of the electromechanical data storage device.

7. The computer-implemented method of claim 1, wherein selecting an electromechanical storage device is based at least on an addressability of one or more failure modes or a probability of failure inherent to the electromechanical data storage device.

8. A computer-implemented method for optimizing data storage, comprising:

determining, by one or more computer systems, one or more modes of failure of a hardware data storage device;

designating a plurality of fault zones based at least in part on the one or more modes of failure;

responding, by one or more computer systems, to data write requests by storing the data associated with the request within a specified zone among the plurality of fault zones, each of the plurality of fault zones comprising a subset of a plurality of data storage locations of the hardware data storage device, each of the subset of data storage locations having a reduced failure correlation with other data storage locations within the subset;

at a time after detecting that one or more physical data locations of the plurality of fault zones to be defective, remapping the defective physical data locations to one or more different physical data locations within the specified zone to which the defective physical data location belongs, such that a logical addressing scheme referencing the defective physical data locations is remapped to instead reference the one or more different physical data locations within the specified zone.

9. The computer-implemented method of claim 8, wherein:

the hardware data storage device is a hard disk drive; and

the plurality of data storage locations are located on platter media of the hard disk drive.

10. The computer-implemented method of claim 9, wherein each fault zone corresponds to a single side of a platter medium.

11. The computer-implemented method of claim 8, wherein the data storage locations within a given fault zone each have a probability of disablement due to malfunction of a common data retrieval mechanism, the probability being within a predetermined threshold of the probability of disablement of the remaining data storage locations within the given fault zone.

12. The computer-implemented method of claim 11, wherein the common data retrieval mechanism is an actuator of a hard disk drive.

13. The computer-implemented method of claim 8, wherein the specified zone is specified by a requesting entity by including an identification of the specified zone in the data write request.

14. A data storage system, comprising:

one or more processors; and

memory, including instructions executable by the one or more processors to configure the data storage system to at least;

select a subset of physical data storage locations based at least in part on a probability of failure inherent to the data storage system;

determining one or more modes of failure of the subset of the physical data storage locations;

designating a placement zone based at least in part on the one or more modes of failure;

map the subset of physical data storage locations of the data storage system to the placement zone, the subset of physical data storage locations sharing a common failure mode that differs in at least one aspect from that of physical data storage locations outside the subset;

service data storage requests by committing data associated with the requests to the placement zone; and

process remap requests for physical data storage locations in the placement zone by remapping the physical data storage locations associated with the request to one or more different physical data storage locations within the placement zone, such that a logical addressing scheme referencing the physical data locations is remapped to instead reference the one or more different physical data locations within the placement zone designated.

15. The data storage system of claim 14, wherein the data storage system is a hard disk drive.

16. The data storage system of claim 15, wherein the placement zone corresponds to a single platter side of the hard disk drive.

17. The data storage system of claim 15, wherein the common failure mode is a head crash event upon a first platter side.

18. The data storage system of claim 17, wherein the aspect in which the common failure mode differs is a head crash event on a platter side that differs from the first platter side.

19. The data storage system of claim 14, wherein the remap requests are initiated by the data storage system in response to receiving information that one or more physical data storage locations are defective.

20. One or more non-transitory computer-readable storage media having collectively stored thereon executable instructions that, when executed by one or more processors of a computer system, cause the computer system to at least:

determine one or more modes of failure of one or more persistent storage media;

generate, from a plurality of storage media locations of the one or more persistent storage media operatively associated with the storage media based at least in part on the one or more modes of failure, the plurality of storage media locations sharing a common failure type or a data durability zone;

when the data durability zone is specified in a data storage request, store data in the data durability zone; and

at a time after receiving information indicating a failure of one or more storage media locations within the data durability zone, remap the failed storage media locations to one or more different storage media locations within the data durability zone, such that a logical addressing scheme referencing the failed storage media locations is remapped to instead reference the one or more different storage media locations within the data durability zone.

21. The one or more non-transitory computer-readable storage media of claim 20, wherein the instructions cause the computer system to further provide a programmatic interface that enables customers of the computer system to at least programmatically provide the data storage request to the computer system.

22. The one or more non-transitory computer-readable storage media of claim 20, wherein the storage media are platter media of a hard disk drive.

23. The one or more non-transitory computer-readable storage media of claim 22, wherein the data durability zone corresponds to a single platter side of the platter media.

24. The one or more non-transitory computer-readable storage media of claim 20, wherein the information is received from the computer system.

25. The one or more non-transitory computer-readable storage media of claim 20, wherein the stored data is erasure coded by the computer system as part of a data reliability enhancement scheme.