Active-active metro-cluster scale-out for unified data path architecture

• US 9,430,480 B1
• Filed: 12/31/2013
• Issued: 08/30/2016
• Est. Priority Date: 12/31/2013
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Bono, Jean-Pierre
• Primary Examiner(s): Choi, Yuk Ting
• Application Number:
  US14/145,351
• Time in Patent Office: 973 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 11/1076 Parity data used in redundant arrays of independent storages, e.g. in RAID systems
• CPC Class Codes: G06F 11/14 Error detection or correction of the data by redundancy in operation G06F11/16 takes precedence
• CPC Class Codes: G06F 11/2094 Redundant storage or storage space G06F11/2056 takes precedence
• CPC Class Codes: G06F 11/2097 maintaining the standby controller/processing unit updated initialisation or re-synchronisation thereof G06F11/1658 and subgroups
• CPC Class Codes: G06F 3/06 Digital input from, or digital output to, record carriers , e.g. RAID, emulated record carriers or networked record carriers
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0619 in relation to data integrity, e.g. data losses, bit errors
• CPC Class Codes: G06F 3/0635 by changing the path, e.g. traffic rerouting, path reconfiguration
• CPC Class Codes: G06F 3/0665 at area level, e.g. provisioning of virtual or logical volumes
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]

Abstract

A technique for providing active-active access to pooled block-based objects and pooled file-based objects by multiple storage processors over distance in a metro-cluster includes virtualizing locally attached LUNs and making the virtualized LUNs available to all the storage processors in the metro-cluster. Storage pools and internal file systems operate on each of the storage processors and coordinate with one another to maintain consistent storage allocation and a consistent file system image across all storage processors in the metro-cluster. The resulting arrangement allows storage processors in the metro-cluster to access both block-based data objects and file-based data objects in an active-active manner, while maintaining consistency across all storage processors and allowing customers to benefit from the advantages of pooled data objects.

Claims

1. A method of processing IO requests in a data storage metro-cluster, comprising:

virtualizing a set of LUNs (logical unit numbers) to make the LUNs accessible by any of multiple storage processors within the data storage metro-cluster;

operating, on each of the storage processors, a clustered storage pool, the clustered storage pool deriving storage units from the set of virtualized LUNs for provisioning to data objects and maintaining consistency with the clustered storage pool on each of the other storage processors such that changes to a clustered storage pool on one storage processor are reflected in the clustered storage pools of all storage processors;

operating, on each of the storage processors, a clustered file system presenting a single namespace shared among the storage processors, the clustered file system provisioned from storage units of the clustered storage pool and storing a data object realized in a form of a file of the clustered file system; and

in response to receiving IO requests, providing read and write access to the data object by the storage processors of the metro-cluster in an active-active arrangement,

wherein the multiple storage processors include a first storage processor at a first location and a second storage processor at a second location, the second location being remote from the first location, wherein maintaining consistency with the clustered storage pool on each of the other storage processors includes mirroring data over distance from the first storage processor at the first location to the second storage processor at the second location, and wherein virtualizing the set of LUNs causes a locally-attached LUN of the first storage processor at the first location to appear as a local LUN of the second storage processor at the second location.

2. A method as in claim 1, wherein the data object is one of (i) a LUN (Logical Unit Number), (ii) a host file system, and a (iii) vVOL (Virtual Volume), realized in the form of the file of the clustered file system.

3. A method as in claim 2, further comprising maintaining cache coherency among data caching of data specified in IO requests received by the storage processors on each of the storage processors of the data storage metro-cluster.

4. A method as in claim 3, wherein, for each of the storage processors, (i) operating the clustered pool and operating the clustered file system are performed in a first container running on the storage processor, and (ii) wherein virtualizing the LUNs and maintaining cache coherency are performed in a second container running on the storage processor, wherein operations in the first container communicate with operations in the second container using IPC (Inter Process Communication).

5. A method as in claim 4, further comprising, for each of the storage processors, performing write-through caching operations in the first container running on the storage processor.

6. A method as in claim 5, further comprising, for each of the storage processors, running an IO stack backend in a third container running on the storage processor, the IO stack backend performing write-back caching of data specified in IO requests received by the storage processor, wherein operations in the third container communicate with operations in the second container using IPC.

7. A method as in claim 2, wherein the data object is a LUN, and wherein the method further comprises:

receiving, at a first storage processor of the storage processors of the metro-cluster, a first IO request in a block-based protocol specifying first data to be written to the LUN;

receiving, at a second storage processor of the storage processors of the metro-cluster, a second IO request in the block-based protocol specifying second data to be written to the LUN;

in response to receiving the first IO request, the first storage processor (i) locating the file in the clustered file system where the LUN is stored and (ii) directing a write operation to write the data specified in the first IO request to the file in the clustered file system; and

in response to receiving the second IO request, the second storage processor (i) locating the file in the clustered file system where the LUN is stored and (ii) directing a write operation to write the data specified in the second IO request to the file in the clustered file system.

8. A method as in claim 2, wherein the data object is a host file system, and wherein the method further comprises operating the host file system in a clustered arrangement such that the host file system presents a single namespace shared among the storage processors of the metro-cluster.

9. A method as in claim 8, wherein the method further comprises:

receiving, at a first storage processor of the storage processors of the metro-cluster, a first IO request in a file-based protocol specifying first data to be written to the host file system;

receiving, at a second storage processor of the storage processors of the metro-cluster, a second IO request in the file-based protocol specifying second data to be written to the host file system;

in response to receiving the first IO request, the first storage processor (i) locating the file in the clustered file system where the host file system is stored and (ii) directing a write operation to write the data specified in the first IO request to the file in the clustered file system; and

in response to receiving the second IO request, the second storage processor (i) locating the file in the clustered file system where the host file system is stored and (ii) directing a write operation to write the data specified in the second IO request to the file in the clustered file system.

10. A data storage metro-cluster comprising multiple storage processors connected together over a network, each of the storage processors including a set of processing units and memory, the memory coupled to the set of processing units and storing executable instructions, which when executed on the respective processing units cause the storage processors to:

virtualize a set of LUNs (logical unit numbers) to make the LUNs accessible by any of multiple storage processors within the data storage metro-cluster;

operate, on each of the storage processors, a clustered storage pool, the clustered storage pool constructed and arranged to (i) derive storage units from the set of virtualized LUNs, (ii) provision storage units to data objects and (iii) maintain consistency with the clustered storage pool on each of the other storage processors such that changes to a clustered storage pool on one storage processor are reflected in the clustered storage pools of all storage processors;

operate, on each of the storage processors, a clustered file system presenting a single namespace shared among the storage processors, the clustered file system provisioned from storage units of the clustered storage pool and storing a data object realized in a form of a file of the clustered file system; and

in response to receiving IO requests, provide read and write access to the data object by the storage processors of the metro-cluster in an active-active arrangement,

wherein the multiple storage processors include a first storage processor at a first location and a second storage processor at a second location, the second location being remote from the first location, wherein the clustered storage pool constructed and arranged to maintain consistency with the clustered storage pool on each of the other storage processors is further constructed and arranged to mirror data over distance from the first storage processor at the first location to the second storage processor at the second location, and wherein virtualizing the set of LUNs is to cause a locally-attached LUN of the first storage processor at the first location to appear as a local LUN of the second storage processor at the second location.

11. A data storage metro-cluster as in claim 10, wherein the data object is one of (i) a LUN (Logical Unit Number), (ii) a host file system, and a (iii) vVOL (Virtual Volume), realized in the form of the file of the clustered file system.

12. A data storage metro-cluster as in claim 11, wherein the executable instructions further cause each of the storage processors to maintain cache coherency among data caching of data specified in IO requests received by the storage processors on each of the storage processors of the data storage metro-cluster.

13. A data storage metro-cluster as in claim 12, wherein the executable instructions further cause each of the storage processors to (i) operate the clustered pool and the clustered file system in a first container running on the respective storage processor, and (ii) virtualize the LUNs and maintain cache coherency in a second container running on the respective storage processor, wherein operations in the first container communicate with operations in the second container using IPC (Inter Process Communication).

14. A data storage metro-cluster as in claim 13, further comprising, for each of the storage processors, a write-through cache in the first container on the storage processor.

15. A data storage metro-cluster as in claim 14, further comprising, for each of the storage processors, an IO stack backend in a third container on the storage processor, the IO stack backend constructed and arranged to perform write-back caching of data specified in IO requests received by the storage processor, wherein the third container is constructed and arranged to communicate with the second container using IPC.

16. A computer program product having a non-transitory computer-readable medium including instructions which, when executed by control circuitry of a data storage system, cause the control circuitry to perform a method of processing IO requests in a data storage metro-cluster, the method comprising:

virtualizing a set of LUNs (logical unit numbers) to make the LUNs accessible by any of multiple storage processors within the data storage metro-cluster;

operating, on each of the storage processors, a clustered storage pool, the clustered storage pool (i) deriving storage units from the set of virtualized LUNs, (ii) provisioning units of storage to data objects and (iii) maintaining consistency with the clustered storage pool on each of the other storage processors such that changes to a clustered storage pool on one storage processor are reflected in the clustered storage pools of all storage processors;

operating, on each of the storage processors, a clustered file system presenting a single namespace shared among the storage processors, the clustered file system provisioned from storage units of the clustered storage pool and storing a data object realized in a form of a file of the clustered file system; and

in response to receiving IO requests, providing read and write access to the data object by the storage processors of the metro-cluster in an active-active arrangement,

wherein the multiple storage processors include a first storage processor at a first location and a second storage processor at a second location, the second location being remote from the first location, wherein maintaining consistency with the clustered storage pool on each of the other storage processors includes mirroring data over distance from the first storage processor at the first location to the second storage processor at the second location, and wherein virtualizing the set of LUNs causes a locally-attached LUN of the first storage processor at the first location to appear as a local LUN of the second storage processor at the second location.

17. A computer program product as in claim 16, wherein the data object is one of (i) a LUN (Logical Unit Number), (ii) a host file system, and a (iii) vVOL (Virtual Volume), realized in the form of the file of the clustered file system.

18. A computer program product as in claim 17, wherein the data object is a LUN, and wherein the method further comprises:

receiving, at a first storage processor of the storage processors of the metro-cluster, a first IO request in a block-based protocol specifying first data to be written to the LUN;

receiving, at a second storage processor of the storage processors of the metro-cluster, a second IO request in the block-based protocol specifying second data to be written to the LUN;

in response to receiving the first IO request, the first storage processor (i) locating the file in the clustered file system where the LUN is stored and (ii) directing a write operation to write the data specified in the first IO request to the file in the clustered file system; and

in response to receiving the second IO request, the second storage processor (i) locating the file in the clustered file system where the LUN is stored and (ii) directing a write operation to write the data specified in the second IO request to the file in the clustered file system.

19. A computer program product as in claim 17, wherein the data object is a host file system, and wherein the method further comprises operating the host file system in a clustered arrangement such that the host file system presents a single namespace shared among the storage processors of the metro-cluster.

20. A method as in claim 19, wherein the method further comprises:

receiving, at a first storage processor of the storage processors of the metro-cluster, a first IO request in a file-based protocol specifying first data to be written to the host file system;

receiving, at a second storage processor of the storage processors of the metro-cluster, a second IO request in the file-based protocol specifying second data to be written to the host file system;

in response to receiving the first IO request, the first storage processor (i) locating the file in the clustered file system where the host file system is stored and (ii) directing a write operation to write the data specified in the first IO request to the file in the clustered file system; and

in response to receiving the second IO request, the second storage processor (i) locating the file in the clustered file system where the host file system is stored and (ii) directing a write operation to write the data specified in the second IO request to the file in the clustered file system.