Great research starts with great data.

Learn More
More >
Patent Analysis of

Data object recovery for storage systems

Updated Time 12 June 2019

Patent Registration Data

Publication Number

US10152376

Application Number

US15/398826

Application Date

05 January 2017

Publication Date

11 December 2018

Current Assignee

EMC IP HOLDING COMPANY LLC

Original Assignee (Applicant)

EMC IP HOLDING COMPANY LLC

International Classification

H03M13/00,G06F11/10,H03M13/37,H03M13/15

Cooperative Classification

G06F11/1076,H03M13/3761,H03M13/1515

Inventor

DANILOV, MIKHAIL,BUINOV, KONSTANTIN,FOMIN, ANDREY,KURILOV, ANDREY,TRUSOV, MAXIM

Patent Images

This patent contains figures and images illustrating the invention and its embodiment.

US10152376 Data object recovery storage 1 US10152376 Data object recovery storage 2 US10152376 Data object recovery storage 3
See all images <>

Abstract

A method comprising: receiving an I/O request for object data; determining one or more data fragments wherein the object data is stored; determining that one or more of the data fragments are unavailable; determining, from within the one or more unavailable data fragments, a set of slices storing the object data, each slice comprising k small data fragments and m coded fragments; for each slice, retrieving at least k small data and coded fragments within the slice from storage; and recovering a segment of the object data using the retrieved small data and coded fragments. A related system and computer program product are also described.

Read more

Claims

1. A method for retrieving data in a storage system, the method comprising:

receiving, at the storage system, an I/O request for object data, the I/O request being received from a client device; identifying one or more data fragments where the object data is stored; determining, by the storage system, that a first one of the data fragments is unavailable due to a failure of a node in the storage system, the first data fragment containing a plurality of first small data fragments that constitute a first portion of the object data; retrieving, a plurality of second small data fragments and a plurality of coded fragments, the plurality of second small data fragments being generated based on the plurality of second small data fragments and a coding matrix, the plurality of second small data fragments and the plurality of coded fragments being part of a slice that is associated with the first data fragment; recovering, by the storage system, the first portion of the object data by using the plurality of second small data fragments and the plurality of coded fragments; and completing, by the storage system, the I/O request for object data based on the recovered first portion of the object data.

2. The method of claim 1, further comprising:

determining, that a second one of the data fragments is available; retrieving the second data fragment; and extracting a second portion of the object data from the second data fragment, wherein the I/O request is completed further based on the extracted second portion of the object data.

3. The method of claim 1, wherein recovering the first portion of the object data comprises performing erasure decoding on the slice.

4. The method of claim 1, wherein the plurality of second small data fragments is retrieved from two or more different nodes within the storage system.

5. The method of claim 1, wherein each of the data fragments is greater than 100 megabytes and each of the first small data fragments is less than 1 kilobyte.

6. The method of claim 5, wherein each of the first small data fragments is a half-byte.

7. A storage system comprising:

one or more processors; a volatile memory; anda non-volatile memory storing computer program code that when executed on the processor causes execution across the one or more processors of a process for retrieving data comprising the operations of:

receiving an I/O request for object data, the I/O request being received from a client device; identifying one or more data fragments where the object data is stored; determining that a first one of the data fragments is unavailable due to a failure of a node in the storage system, the first data fragment containing a plurality of first small data fragments that constitute a first portion of the object data; retrieving a plurality of second small data fragments and a plurality of coded fragments, the plurality of second small data fragments being generated based on the plurality of second small data fragments and a coding matrix, the plurality of second small data fragments and the plurality of coded fragments being part of a slice that is associated with the first data fragment; recovering the first portion of the object data by using the plurality of second small data fragments and the plurality of coded fragments; and completing the I/O request for object data based on the recovered first portion of the object data.

8. The storage system of claim 7, wherein the computer program code causes execution across the one or more processors of the process further operable to perform the operations of:

determining that a second one of the data fragments is available; retrieving the second data fragment; and extracting a second portion of the object data from the retrieved second data fragment, wherein the I/O request is completed further based on the extracted second portion of the object data.

9. The storage system of claim 7, wherein recovering the first portion of the object data comprises performing erasure decoding on the slice.

10. The storage system of claim 7, wherein each of the data fragments is greater than 100 megabytes and each of the first small data fragments is less than 1 kilobyte.

11. The storage system of claim 10, wherein each of the first small data fragments is a half-byte.

12. A computer program product tangibly embodied in a non-transitory computer-readable medium, the computer-readable medium storing program instructions which when executed by at least one processor that is part of a storage system cause the at least one processor to:

receive an I/O request for object data that is transmitted to the storage system by a client device; identify one or more data fragments where the object data is stored; determine that a first one of the data fragments is unavailable due to a failure of a node in the storage system, the first data fragment containing a plurality of first small data fragments that constitute a first portion of the object data; retrieving, a plurality of second small data fragments and a plurality of coded fragments, the plurality of second small data fragments being generated based on the plurality of second small data fragments and a coding matrix, the plurality of second small data fragments and the plurality of coded fragments being part of a slice that is associated with the first data fragment; recover the first portion of the object data by using the plurality of second small data fragments and the plurality of coded fragments; and complete the I/O request for object data based on the recovered first portion of the object data.

13. The computer program product of claim 12, wherein the program instructions are further executable to:

determine that a second one of the data fragments is available; retrieve the second data fragment; and extract a second portion of the object data from the retrieved second data fragment, wherein the I/O request is completed further based on the extracted second portion of the object data.

14. The computer program product of claim 12 wherein recovering the first portion of the object data comprises performing erasure decoding on the retrieved slices.

15. The computer program product of claim 12, wherein the plurality of second small data fragments is retrieved from two or more different nodes within the storage system.

16. The computer program product of claim 12, wherein each of the data fragments is greater than 100 megabytes and each of the first small data fragments is less than 1 kilobyte.

17. The computer program product of claim 16, wherein each of the first small data fragments is a half-byte.

Read more

Claim Tree

  • 1
    1. A method for retrieving data in a storage system, the method comprising:
    • receiving, at the storage system, an I/O request for object data, the I/O request being received from a client device
    • identifying one or more data fragments where the object data is stored
    • determining, by the storage system, that a first one of the data fragments is unavailable due to a failure of a node in the storage system, the first data fragment containing a plurality of first small data fragments that constitute a first portion of the object data
    • retrieving, a plurality of second small data fragments and a plurality of coded fragments, the plurality of second small data fragments being generated based on the plurality of second small data fragments and a coding matrix, the plurality of second small data fragments and the plurality of coded fragments being part of a slice that is associated with the first data fragment
    • recovering, by the storage system, the first portion of the object data by using the plurality of second small data fragments and the plurality of coded fragments
    • and completing, by the storage system, the I/O request for object data based on the recovered first portion of the object data.
    • 2. The method of claim 1, further comprising:
      • determining, that a second one of the data fragments is available
      • retrieving the second data fragment
      • and extracting a second portion of the object data from the second data fragment, wherein the I/O request is completed further based on the extracted second portion of the object data.
    • 3. The method of claim 1, wherein
      • recovering the first portion of the object data comprises
    • 4. The method of claim 1, wherein
      • the plurality of second small data fragments is retrieved from two or more different nodes within the storage system.
    • 5. The method of claim 1, wherein
      • each of the data fragments is greater than 100 megabytes and each of the first small data fragments is less than 1 kilobyte.
  • 7
    7. A storage system comprising:
    • one or more processors
    • a volatile memory
    • anda non-volatile memory storing computer program code that when executed on the processor causes execution across the one or more processors of a process for retrieving data comprising the operations of: receiving an I/O request for object data, the I/O request being received from a client device
    • identifying one or more data fragments where the object data is stored
    • determining that a first one of the data fragments is unavailable due to a failure of a node in the storage system, the first data fragment containing a plurality of first small data fragments that constitute a first portion of the object data
    • retrieving a plurality of second small data fragments and a plurality of coded fragments, the plurality of second small data fragments being generated based on the plurality of second small data fragments and a coding matrix, the plurality of second small data fragments and the plurality of coded fragments being part of a slice that is associated with the first data fragment
    • recovering the first portion of the object data by using the plurality of second small data fragments and the plurality of coded fragments
    • and completing the I/O request for object data based on the recovered first portion of the object data.
    • 8. The storage system of claim 7, wherein
      • the computer program code causes execution across the one or more processors of the process further operable to perform the operations of: determining that a second one of the data fragments is available; retrieving the second data fragment; and extracting a second portion of the object data from the retrieved second data fragment, wherein
    • 9. The storage system of claim 7, wherein
      • recovering the first portion of the object data comprises
    • 10. The storage system of claim 7, wherein
      • each of the data fragments is greater than 100 megabytes and each of the first small data fragments is less than 1 kilobyte.
  • 12
    12. A computer program product tangibly embodied in a non-transitory computer-readable medium, the computer-readable medium storing program instructions which when executed by at least one processor that is part of a storage system cause the at least one processor to: receive an I/O request for object data that is transmitted to the storage system by a client device; identify one or more data fragments where the object data is stored; determine that a first one of the data fragments is unavailable due to a failure of a node in the storage system, the first data fragment containing
    • a plurality of first small data fragments that constitute a first portion of the object data
    • retrieving, a plurality of second small data fragments and a plurality of coded fragments, the plurality of second small data fragments being generated based on the plurality of second small data fragments and a coding matrix, the plurality of second small data fragments and the plurality of coded fragments being part of a slice that is associated with the first data fragment
    • recover the first portion of the object data by using the plurality of second small data fragments and the plurality of coded fragments
    • and complete the I/O request for object data based on the recovered first portion of the object data.
    • 13. The computer program product of claim 12, wherein
      • the program instructions are further executable to: determine that a second one of the data fragments is available; retrieve the second data fragment; and extract a second portion of the object data from the retrieved second data fragment, wherein
    • 14. The computer program product of claim 12 wherein
      • recovering the first portion of the object data comprises
    • 15. The computer program product of claim 12, wherein
      • the plurality of second small data fragments is retrieved from two or more different nodes within the storage system.
    • 16. The computer program product of claim 12, wherein
      • each of the data fragments is greater than 100 megabytes and each of the first small data fragments is less than 1 kilobyte.
See all independent claims <>

Description

CROSS REFERENCE TO RELATED APPLICATIONS

This application claims priority to Russian Patent Application number 2016125855, filed Jun. 29, 2016, and entitled “DATA OBJECT RECOVERY FOR STORAGE SYSTEMS,” which is incorporated herein by reference in its entirety.

BACKGROUND

Distributed storage systems may provide a wide range of storage services, while achieving high scalability, availability, and serviceability. An example of a distributed storage system is Elastic Cloud Storage (ECS) from EMC Corporation of Hopkinton, Massachusetts.

Erasure coding is a coding technique originally created as a forward error correction method for binary erasure channels. More recently, erasure coding has been used to protect data within storage systems, including distributed storage systems. If a storage node fails, data stored thereon may be recovered using coded data stored on other nodes. Existing erasure coding implementations use resource-demanding data recovery techniques that can negatively impact user I/O.

SUMMARY

Described herein are embodiments of systems and techniques to reduce the impact of data recovery on I/O. Various embodiments use pointed data recovery that can reduce I/O errors during data recovery after a hardware failure.

According to one aspect of the disclosure, a method comprises: receiving an I/O request for object data; determining one or more data fragments wherein the object data is stored; determining that one or more of the data fragments are unavailable; determining, from within the one or more unavailable data fragments, a set of slices storing the object data, each slice comprising k small data fragments and m coded fragments; for each slice, retrieving at least k small data and coded fragments within the slice from storage; and recovering a segment of the object data using the retrieved small data and coded fragments.

In some embodiment, the method further comprises determining that one or more of the data fragments is available, retrieving the one or more available data fragments, extracting a segment of the object data from the retrieved available data fragments, and generating the object data using the recovered segment of the object data and the retrieved segment of the object data. In many embodiments, recovering the segment of the object data using the retrieved small data and coded fragments comprises performing erasure decoding on each slice using the at least k small data and coded fragments within the slice. In certain embodiments, for each slice, retrieving the at least k small data and coded fragments within the slice from storage comprises retrieving at least k small data and coded fragments from two or more different nodes within a storage cluster. In particular embodiments, each of the data fragments is greater than 100 megabytes and each of the small data fragments is less than 1 kilobyte. In one embodiment, each of the small data fragments is a half-byte.

According to another aspect of the disclosure, a system comprises one or more processors; a volatile memory; and a non-volatile memory storing computer program code that when executed on the processor causes execution across the one or more processors of a process operable to perform embodiments of the method described hereinabove.

According to yet another aspect of the disclosure, a computer program product tangibly embodied in a non-transitory computer-readable medium, the computer-readable medium storing program instructions that are executable to perform embodiments of the method described hereinabove.

BRIEF DESCRIPTION OF THE DRAWINGS

The concepts, structures, and techniques sought to be protected herein may be more fully understood from the following detailed description of the drawings, in which:

FIG. 1 is a block diagram of a storage system, in accordance with an embodiment of the disclosure;

FIG. 1A is a block diagram of a storage node that may form a part of a storage system, in accordance with an embodiment of the disclosure;

FIG. 2 is a diagram of a storage system using erasure coding to protect data, in accordance with an embodiment of the disclosure;

FIG. 3 is a diagram of a data storage chunk, in accordance with an embodiment of the disclosure;

FIG. 3A is a diagram of a data storage chunk having unavailable data fragments, in accordance with an embodiment of the disclosure;

FIG. 4 is a diagram of another data storage chunk, in accordance with an embodiment of the disclosure;

FIG. 4A is a diagram of a data storage chunk comprising a plurality of slices, in accordance with an embodiment of the disclosure;

FIG. 5 is a flow diagram showing a method that may be used within the systems of FIGS. 1 and 1A, in accordance with an embodiment of the disclosure; and

FIG. 6 is a schematic representation of a computer for use with the systems of FIGS. 1 and 1A, in accordance with an embodiment of the disclosure.

The drawings are not necessarily to scale, or inclusive of all elements of a system, emphasis instead generally being placed upon illustrating the concepts, structures, and techniques sought to be protected herein.

DETAILED DESCRIPTION

Before describing embodiments of the structures and techniques sought to be protected herein, some terms are explained. In certain embodiments, the term “storage system” may encompass private or public cloud computing systems for storing data as well as systems for storing data comprising virtual infrastructure and those not comprising virtual infrastructure. In some embodiments, the term “I/O request” may refer to a request to read and/or write data. In many embodiments, the terms “client” and “user” may refer to any person, system, or other entity that may send I/O requests to a storage system.

In certain embodiments, the term “storage device” may refer to any non-volatile memory (NVM) device, including hard disk drives (HDDs), flash devices (e.g., NAND flash devices), and next generation NVM devices, any of which may be accessed locally and/or remotely (e.g., via a storage attached network (SAN)). In some embodiments, the term “storage device” may also refer to a storage array comprising one or more storage devices.

FIG. 1 shows a storage system 100, in accordance with an embodiment of the disclosure. The storage system 100 includes one or more clients 102 in communication with a storage cluster 104 via a network 103. The storage cluster 104 includes one or more storage nodes 106a . . . 106n (generally denoted 106). In general operation, clients 102 issue requests to the storage cluster 104 to read and write data. A given request is received by any available storage node 106. The receiving node 106 processes the request locally and/or delegates request processing to one or more other nodes 106.

In some embodiments, the network may include any suitable type of communication network or combination thereof, including networks using protocols such as Ethernet, Internet Small Computer System Interface (iSCSI), Fibre Channel (FC), and/or wireless protocols. In certain embodiments, clients may include user applications, application servers, data management tools, and/or testing systems. In particular embodiments, a storage node may be the same as or similar to an embodiment shown in FIG. 1A and described below in conjunction therewith. In some embodiments, the storage cluster may utilize Elastic Cloud Storage (ECS) from EMC Corporation of Hopkinton, Massachusetts.

FIG. 1A shows an illustrative storage node 106′, in accordance with an embodiment of the disclosure. The storage node 106′ includes one or more services 108 and one or more storage devices 110. The services 108 include: an authentication service 108a to authenticate requests from clients 102; storage API services 108b to parse and interpret requests from clients 102; a storage chunk management service 108c to facilitate storage chunk allocation/reclamation for different storage system needs and monitor storage chunk health and usage; a storage server management service 108d to manage available storage devices capacity and to track storage devices states; and a storage server service 108e to interface with the storage devices 110.

In one embodiment, the storage node may include a processor and a non-volatile memory storing computer program code that when executed on the processor causes the processor to execute processes operable to perform functions of the services.

In some embodiments, storage devices may comprise one or more physical and/or logical storage devices attached to the storage node. In certain embodiments, storage devices may be provided as a storage array. In particular embodiments, storage devices may be provided as VNX or Symmetrix VMAX, which are available from EMC Corporation of Hopkinton, Massachusetts.

In some embodiments, the storage devices may be partitioned into sets of fixed-sized blocks referred to as “chunks” within which user data (e.g., object data) is stored. In certain embodiments, the size of a chunk is 128 MB. A given object may be fragmented across multiple chunks, and a given chunk may store fragments of multiple different objects. In various embodiments, new data may be appended to a chunk, but existing chunk data cannot be modified. In many embodiments, when a chunk becomes full, it may be marked as “sealed.” Sealed chunks are immutable.

In various embodiments, the storage node 106′ may implement processing described below in conjunction with the embodiments of FIGS. 2-5 to protect against data loss. In particular embodiments, the storage chunk management service 108c may be configured to perform such processing.

Referring to FIG. 2, in some embodiments, a storage system may use erasure coding to protect against data loss. In the embodiment shown, a storage system 200 performs erasure coding on chunks that have been sealed. A sealed chunk is split into k data fragments D1, D2, . . . , Dk and, using coding techniques described below, m coded fragments C1, C2, . . . , Cm are generated from the k fragments. The coded fragments C include redundant information that can be used to regenerate (or “restore”) data fragments D in certain situations.

In some embodiments, the storage system may use a coding scheme that allows it to tolerate the loss of any m data fragments within a chunk. In the embodiment of FIG. 2, k=12 and m=4, although other values of k and m can be used.

In certain embodiments, a chunk may be split into k equal size data fragments D1, D2, . . . , Dk, with padding or other data complement being added as needed to ensure the data fragments are of equal size.

Referring again to FIG. 2, the storage system 200 is a distributed storage system having a plurality of storage nodes 201-216, any of which may be the same as or similar to embodiments described above in conjunction with FIG. 1A. The data fragments D1, D2, . . . , Dk and coded fragments C1, C2, . . . , Cm are stored across multiple storage nodes 201-216. The storage system 200 includes sixteen (16) storage nodes 201-216, with each node 201-216 storing exactly one data fragment D or coded fragment C.

In other embodiments, a give storage node may store multiple data fragments and/or multiple coded fragments (e.g., if k+m is greater than the number of available nodes). In still other embodiments, some storage nodes may not store any data fragments or coded fragments (e.g., if k+m is less than the number of available nodes). In various embodiments, a storage system may include fewer than or more than sixteen (16) storage nodes.

In some embodiments, tables or other suitable data structures may be maintained to track the location of data fragments D and coded fragments C within the storage system 200. In particular embodiments, a mapping is maintained between object keys and one or more storage locations where corresponding object data is stored within the storage system 200. In certain embodiments, a first table is maintained to map object keys to chunks, and a second table is maintained to track the location of chunk data and coded fragments.

Referring again to FIG. 2, if a data fragment D is lost (e.g., due to a node failure, a storage device failure, or data corruption), the lost data fragment can be recovered using other data fragments D and coded fragments C. At least k unique fragments—either data fragments or coded fragments—within the same chunk are required to recover a lost data fragment via decoding. Thus, the storage system 200 is able to tolerate the loss of any m fragments within a given chunk.

In various embodiments, when an I/O request is received for some object data (e.g., from a client 102 in FIG. 1), the storage system 100 may determine which chunk(s) store the requested object data and where the corresponding chunk data fragments are stored (e.g., which nodes 201-216 the data fragments are stored on). Referring back to FIG. 2, the system 200 may then attempt to read one or more object segments from the chunk data fragments D. If any of the chunk data fragments are unavailable (e.g., due to a storage node failure), the system 200 may attempt to recover the unavailable data fragments (or portions thereof) using other data and coded fragments within the same chunk. This type of recover is referred to here as “on-demand recovery.” In some embodiments, on-demand recover may be performed as part of the I/O request processing. In various embodiments, it is a goal to improve the efficiency of on-demand data recovery to reduce I/O request latency and to reduce the probability of I/O errors (e.g., I/O errors caused by server and/or network timeouts).

Referring back to FIG. 2, storage system 200 may perform bulk (or “batch”) data recovery when a storage node 106 fails or otherwise becomes unavailable. Bulk data recover may occur in the background (i.e., separate from I/O request processing) and may be used in conjunction with on-demand data recovery (i.e., the two techniques may be complementary). In particular embodiments, the storage system 200 may maintain a list of chunks data fragments that require recovery and process the list in the background. In various embodiments, it is a goal to reduce the impact of background bulk data recovery on user I/O request processing.

In some embodiments, a matrix-based Reed-Solomon erasure coding technique may be used for data protection within a storage system. In certain embodiments, a storage chunk may be split into k data fragments D1, D2, . . . , Dk and the data fragments D may be arranged as a column vector (“data vector”). In some embodiments, m coded fragments C1, C2, . . . , Cm may be generated by multiplying an m×k matrix of coding coefficients Xij (referred to herein as a “coding matrix”) by the k-element data vector. In some embodiments, the decoding matrix used may depend on the specific data and coded fragments that are available and/or on the specific data fragments to be recovered. In many embodiments, a second m×k matrix of coefficients Yij (referred to herein as a “decoding matrix”) can be used to recover unavailable data fragments, as described below. In certain embodiments, the coefficients of the coding matrix X and/or the decoding matrix Y may be selected using known erasure coding techniques and/or based upon the specific erasure coding algorithm used. In some embodiments, the coefficients of the coding matrix X and/or the decoding matrix Y may be selected such that the storage system can tolerate the loss of m data fragments within a chunk.

Referring to FIGS. 3 and 3A, in various embodiments, it may be possible to recover requested data (e.g., object data) without recovering all unavailable data fragments within a chunk. Instead, it may be sufficient to recover only those data fragments (or portions thereof) storing the requested data, thereby reducing processing and I/O costs.

FIG. 3 shows an object data 302 stored within an encoded chunk 300 (e.g., a chunk that has been encoded using erasure encoding), according to an embodiment. The encoded chunk 300 includes k data fragments D1, D2, . . . , Dk and m coded fragments C1, C2, . . . , Cm, with k=12 and m=4 in this embodiment. The object data 302 may correspond to a full object or a portion thereof (sometimes referred to as an “object segment”). The object data 302 is stored in the data fragment D7. In some embodiments, the size of object data may be substantially less than the size of a data fragment D. In one embodiment, the size of object data 302 may be about 10 KB whereas the size of a data fragment D may be about 10 MB.

Referring to FIG. 3A, a storage node failure can cause chunk data fragments D1, D5, D7, and D12 to become unavailable, according to an embodiment. If an I/O request arrives for the object data 302—which is stored in unavailable data fragment D7—before bulk data recover recovers data fragment D7, the I/O request triggers on-demand recovery of data fragment D7 (or a portion thereof). In one embodiment, the I/O request may be a partial read that references the object data. In many embodiments, the term “partial read” refers to an I/O request to read only certain segment(s) of an object, not the entire object.

In various embodiments, unavailable data fragments D can be recovered using an erasure decoding process, wherein inputs to the process may include a decoding matrix Y, one or more coded fragments C, and one or more available data fragments.

Referring back to FIG. 3A, the unavailable data fragments D1, D5, D7, and D12 may be recovered using the following equations:

D1=Y1,1D2+Y1,2D3+Y1,3D4Y1,4D6+Y1,5D8+Y1,6D9+Y1,7D10+Y1,8*D11+Y1,9C1+Y1,10C2+Y1,11C3+Y1,12C4

D5=Y2,1D2+Y2,2D3+Y2,3D4+Y2,4D6+Y2,5D8+Y2,6D9+Y2,7D10+Y2,8D11+Y2,9C1+Y2,10C2+Y2+Y2,11C3+Y2,12C4

D7=Y3,1D2+Y3,2D3+Y3,3D4+Y3,4D6+Y3,5D8+Y3,6D9+Y3,7D10+Y3,8D11+Y9C1+Y3,10C2+Y3,11C3+Y3,12C4

D12=Y4,1D2+Y4,2D3+Y4,3D4Y4,4D6+Y4,5D8+Y4,6D9+Y4,7D10+Y4,8D11+Y4,9C1+Y4,10C2+Y4,11C3+Y4,12C4

In some embodiments, decoding a full data fragment (e.g., D7) requires retrieving k fragments (data or coded fragments) within the same chunk, with k=12 in the embodiment of FIG. 3A. In various embodiments, the fragments may be stored across multiple storage nodes and, thus, retrieving a fragment may incur both network and storage I/O costs. In certain embodiments, recovery of a full data fragment D may require retrieving 128 MB of data.

In various embodiments, in order reduce processing and/or I/O costs, on-demand data recovery recovers only those data fragments D within a chunk that are needed to process the I/O request. For example, referring to FIG. 3A, only data fragment D7 may be restored when processing an I/O request for object segment 302.

Referring to FIGS. 4 and 4A, according some embodiments, it may be possible to recover object data without having to retrieve k full data and/or coded fragments. For example, referring to FIG. 3A, object data 302 may be recovered without fully decoding D7 and, thus, without having to read the full contents of data fragments D2, D3, D4, D6, D8, D9, D10, D11 and coded fragments C1, C2, C3, C4. Instead, only a fraction of the available data fragments are read and decoded, which can decreases data recovery time resulting in faster servicing of I/O requests and a decreased probability of failed I/O requests.

In some embodiments, chunk erasure coding/decoding may be performed using units of data that are smaller than the size of a data fragment D.

Referring to FIG. 4, with k=4 and m=2, erasure coding may be performed on 8-byte chunks 400, according to an embodiment. A chunk 400 is divided into four 2-byte data fragments D1, D2, D3, D4 that can be multiplied by a coding matrix of coding coefficients Xij (not shown) to generate two 2-byte coded fragments C1, C2. Each data fragment D is split into multiple parts referred to herein as “small data fragments” and denoted di in the figures. Each two-byte data fragment D is split into four (4) small data fragments di, each small data fragment comprising a half-byte value (or “nibble”). In particular, a first data fragment D1 is split into four (4) small data fragments d1, d2, d4, a second data fragment D2 is split into four small data fragments d5, d6, d7, d8, and so on. In other embodiments, the size of the small data fragments di may be smaller than a half-byte (e.g., 1 bit) or larger than a half-byte (e.g., several bytes).

In various embodiments, so-called “small coded fragments” ci may be generated by multiplying small data fragments di by coding coefficients Xij. Referring to FIG. 4, a first small coded fragment c1 may be calculated as:

C1=X1,1*d1+X1,2*d5+X1,3*d9+X1,4*d13

  • and a fifth small coded fragment c5 may be calculated as:

    C5=X2,1*d1+X2,2*d5+X2,3*d9+X2,4*d13.

In certain embodiments, a plurality of small coded fragments can be combined to form a full coded fragment C. In the example of FIG. 4, small coded fragments c1, c2, c3, c4 may be combined to form a first coded fragment C1, and small coded fragments c5, c6, c7, C8 may be combined to form a second coded fragment C2.

Referring to FIG. 4A, in some embodiments, erasure coding can be implemented as a series of operations on sets of small data fragments di and/or small coded fragments ci. Each set of small data fragments, referred to herein as a “slice”402, includes k small data fragments di and m small coded fragments ci. For example, as shown in FIG. 4A, a first slice 402a may include small data fragments d1, d5, d9, d13, and small coded fragments c1, c5.

In some embodiments, the number of slices (L) per chunk is equal to the size of a data fragment Di divided by the size of a small data fragment di. In the example shown, L=2 bytes/0.5 bytes=4. In certain embodiments, slices may be identified by a sequence number between one (1) and L.

Referring back to FIG. 4A, for a given slice 402, the small coded fragments ci within the slice can be generated using the small data fragments di within the slice and the coding matrix X. Likewise, a small data fragment di can be recovered using small data/coded fragments within the same slice and the decoding matrix Y. In certain embodiments, a small data fragment di can be recovered using k unique small fragments (data or coded fragments) within the same slice.

In various embodiments, a storage system may recover object data by retrieving and decoding one or more slices. In certain embodiments, the number of slices required to be retrieved and/or decoded may depend on the amount of data to be recovered. In one embodiment, 10 KB of data can be recovered by retrieving and processing 120 KB of data within the same set of slices. In certain embodiments, the technique described herein (referred to as “slice-based recovery”) may be used conditionally. In some embodiments, slice-based recovery may be used if the amount of data to be recovered is less than a threshold value: for larger object segments, it may be more efficient (e.g., in terms of I/O and/or processing) to decode full data fragments D.

In some embodiments, decoding can be accelerated using special processor instructions like VPERM for PowerPC and/or PSHUFB for Intel processors.

FIG. 5 is a flow diagram showing illustrative processing in accordance with certain embodiments of the disclosure. In some embodiments, processing may be implemented within one or more nodes 106 of storage system 100 (FIG. 1). Rectangular elements (typified by element 502) herein denoted “processing blocks,” represent computer software instructions or groups of instructions. Diamond shaped elements (typified by element 506) herein denoted “decision blocks,” represent computer software instructions, or groups of instructions, which affect the execution of the computer software instructions represented by the processing blocks.

Alternatively, the processing and decision blocks may represent steps performed by functionally equivalent circuits such as a digital signal processor circuit or an application specific integrated circuit (ASIC). The flow diagrams do not depict the syntax of any particular programming language. Rather, the flow diagrams illustrate the functional information one of ordinary skill in the art requires to fabricate circuits or to generate computer software to perform the processing required of the particular apparatus. It should be noted that many routine program elements, such as initialization of loops and variables and the use of temporary variables are not shown. It will be appreciated by those of ordinary skill in the art that unless otherwise indicated herein, the particular sequence of blocks described is illustrative only and can be varied without departing from the spirit of the concepts, structures, and techniques sought to be protected herein. Thus, unless otherwise stated the blocks described below are unordered meaning that, when possible, the functions represented by the blocks can be performed in any convenient or desirable order.

Referring to FIG. 5, a process 500 begins at block 502, where an I/O request is received for object data. In some embodiments, the I/O request is received from a client 102 (FIG. 1). The I/O request may include a request to read/write a full object or partial object data (e.g., a partial read).

Referring again to FIG. 5, at block 504, one or more chunk data fragments D are determined for the object data (e.g., the data fragments D that store the requested object data). In some embodiments, the object data may be stored across multiple chunks and, thus, determining the one or more chunk data fragments may include determining data fragments from two or more different chunks. In certain embodiments, the I/O request may include an object key and determining the one or more chunk data fragments may include using the object key to determine one or more storage chunks for the requested object data and, within those chunks, one or more data fragments D. In particular embodiments, the data fragments D may be stored across one or more storage nodes of a distributed storage system.

Referring again to FIG. 5, at block 505, the process 500 attempts to retrieve all the data fragments D from storage. In certain embodiments, the process may determine that one or more of the data fragments D are unavailable. For example, the process may determine that a storage node storing a data fragment D is not available and, thus, may avoid attempting to retrieve a data fragment D therefrom. In either case, if a data fragment D cannot be retrieved, it may be treated as unavailable.

Referring back to FIG. 5, at block 506, if all the data fragments D are available (e.g., the were successfully retrieved), then the object data is extracted therefrom and returned (block 508). In some embodiments, the object data may be returned to a client.

Referring again to FIG. 5, at block 506, if one or more of the data fragments D is not available (e.g., due to a failed storage node), then on-demand data recovery is performed. The process 500 supports two different types of on-demand recovery: (1) full data fragment recovery and (2) slice-based recovery. At block 510, a decision may be made as to which type of recovery to perform. In certain embodiments, this decision is based on factors described above in conjunction with FIG. 4A.

Referring again to FIG. 5, if full data fragment recovery is decided, then processing may proceed to block 512 where, for each unavailable data fragment, k unique fragments (either data fragments D or coded fragments C) within the same chunk may be retrieved from storage. In some embodiments, the specific data fragments D and coded fragments C retrieved may depend on the erasure encoding implementation, the encoding matrix X, and/or the decoding matrix Y. For example, using the equations shown above in conjunction with FIG. 3A, unavailable data fragment D7 may be recovered by retrieving data fragments D2, D3, D4, D6, D8, D9, D10, D11 and coded fragments C1, C2, C3, C4.

At block 514, the unavailable data fragments D may be recovered using the retrieved data/coded fragments. In certain embodiments, recovering unavailable data fragments includes multiplying the retrieved data and coded fragments by a decoding matrix Y.

Referring back to FIG. 5, the requested object data may be extracted from the recovered data fragments D (and, possibly, from data fragments successfully retrieved at block 505) and returned (block 508).

Still referring back to FIG. 5, if slice-based data fragment recovery is decided, then processing proceeds to block 518s, where one or more small data fragments di, are determined. The small data fragments di correspond to portions of the unavailable data fragments D where the object data is stored. Thus, the small data fragments di determined at block 518 are referred to as “unavailable small data fragments.” For each unavailable small data fragment di, a corresponding slice is also determined. It is possible that the requested object data is stored within two or more different unavailable data fragments D. In this case, the set of slices to recover may be the union of ranges calculated for each affected object segment. In certain embodiments, the process determines the minimal range of slices needed to recover the requested object data. For example, referring to FIGS. 4 and 4A, if data fragment D1 is unavailable and the object data is stored within small data fragments d2 and d3, then slices 402b and 402c may be determined at block 518.

Referring again to FIG. 5, at block 520, for each unavailable small data fragment, k unique small fragments (either small data fragments di or small coded fragments ci) within the same slice are retrieved from storage. At block 522, the unavailable small data fragments di are recovered using the retrieved small data and coded fragments. In certain embodiments, recovering small data fragments includes arranging the retrieved small data and coded fragments into a column vector and multiplying the column vector by a decoding matrix Y.

Referring back to FIG. 5, the requested object data is extracted from the recovered small data fragments di (and, possibly, from data fragments successfully retrieved at block 505) and returned (block 508).

In some embodiments, both full data fragment recovery and slice-based recovery may be used when processing an I/O request. For example, in some embodiments if multiple data fragments are unavailable for a requested object data, at least one of the unavailable data fragments D may be recovered in full, whereas at least one different unavailable fragment may be partially recovered slice-based recovery.

FIG. 6 shows an illustrative computer or other processing device 600 that can perform at least part of the processing described herein, in accordance with an embodiment of the disclosure. The computer 600 includes a processor 602, a volatile memory 604, a non-volatile memory 606 (e.g., hard disk), an output device 608 and a graphical user interface (GUI) 610 (e.g., a mouse, a keyboard, a display, for example), each of which is coupled together by a bus 618. The non-volatile memory 606 stores computer instructions 612, an operating system 614, and data 616. In one example, the computer instructions 612 are executed by the processor 602 out of volatile memory 604. In one embodiment, an article 620 comprises non-transitory computer-readable instructions.

Processing may be implemented in hardware, software, or a combination of the two. In various embodiments, processing is provided by computer programs executing on programmable computers/machines that each includes a processor, a storage medium or other article of manufacture that is readable by the processor (including volatile and non-volatile memory and/or storage elements), at least one input device, and one or more output devices. Program code may be applied to data entered using an input device to perform processing and to generate output information.

The system can perform processing, at least in part, via a computer program product, (e.g., in a machine-readable storage device), for execution by, or to control the operation of, data processing apparatus (e.g., a programmable processor, a computer, or multiple computers). Each such program may be implemented in a high level procedural or object-oriented programming language to communicate with a computer system. However, the programs may be implemented in assembly or machine language. The language may be a compiled or an interpreted language and it may be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program may be deployed to be executed on one computer or on multiple computers at one site or distributed across multiple sites and interconnected by a communication network. A computer program may be stored on a storage medium or device (e.g., CD-ROM, hard disk, or magnetic diskette) that is readable by a general or special purpose programmable computer for configuring and operating the computer when the storage medium or device is read by the computer. Processing may also be implemented as a machine-readable storage medium, configured with a computer program, where upon execution, instructions in the computer program cause the computer to operate. The program logic may be run on a physical or virtual processor. The program logic may be run across one or more a physical or virtual processors.

Processing may be performed by one or more programmable processors executing one or more computer programs to perform the functions of the system. All or part of the system may be implemented as special purpose logic circuitry (e.g., an FPGA (field programmable gate array) and/or an ASIC (application-specific integrated circuit)).

All references cited herein are hereby incorporated herein by reference in their entirety.

Having described certain embodiments, which serve to illustrate various concepts, structures, and techniques sought to be protected herein, it will be apparent to those of ordinary skill in the art that other embodiments incorporating these concepts, structures, and techniques may be used. Elements of different embodiments described hereinabove may be combined to form other embodiments not specifically set forth above and, further, elements described in the context of a single embodiment may be provided separately or in any suitable sub-combination. Accordingly, it is submitted that scope of protection sought herein should not be limited to the described embodiments but rather should be limited only by the spirit and scope of the following claims.

Read more
PatSnap Solutions

Great research starts with great data.

Use the most comprehensive innovation intelligence platform to maximise ROI on research.

Learn More

Citation

Patents Cited in This Cited by
Title Current Assignee Application Date Publication Date
Method for distributed direct object access storage OATH INC. 25 August 2009 03 March 2011
Erasure Coded Storage Aggregation in Data Centers MICROSOFT TECHNOLOGY LICENSING, LLC 31 July 2009 03 February 2011
Storage of Data In A Distributed Storage System GOOGLE LLC 08 February 2011 11 August 2011
Redundant array of independent clouds CIRTAS SYSTEMS, INC. 20 August 2010 23 February 2012
Distributed file storage and transmission system WAMBO, INC. 02 May 2007 29 May 2008
See full citation <>

More like this

Title Current Assignee Application Date Publication Date
Data storage system and data storage method TSINGHUA UNIVERSITY 09 December 2014 16 June 2016
Distributed data storage method, control device and system HUAWEI TECHNOLOGIES CO., LTD. 17 June 2015 22 December 2016
Fault-tolerant enterprise object storage system for small objects CISCO TECHNOLOGY, INC. 25 April 2017 02 November 2017
Distributed storage system data management and security CLOUD CROWDING CORP. 15 March 2017 21 September 2017
Coding for distributed storage system HUAWEI TECHNOLOGIES CO., LTD. 09 October 2015 15 June 2017
See all similar patents <>

More Patents & Intellectual Property

PatSnap Solutions

PatSnap solutions are used by R&D teams, legal and IP professionals, those in business intelligence and strategic planning roles and by research staff at academic institutions globally.

PatSnap Solutions
Search & Analyze
The widest range of IP search tools makes getting the right answers and asking the right questions easier than ever. One click analysis extracts meaningful information on competitors and technology trends from IP data.
Business Intelligence
Gain powerful insights into future technology changes, market shifts and competitor strategies.
Workflow
Manage IP-related processes across multiple teams and departments with integrated collaboration and workflow tools.
Contact Sales
Clsoe
US10152376 Data object recovery storage 1 US10152376 Data object recovery storage 2 US10152376 Data object recovery storage 3