Data storage system binding virtual volumes to host-specific protocol endpoints

• US 10,891,060 B2
• Filed: 10/31/2018
• Issued: 01/12/2021
• Est. Priority Date: 10/31/2018
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Inventors: Tylik, Dmitry, Koluguri, Anil Kumar, Long, Matthew H., Stotski, Yuri A.
• Primary Examiner(s): Chappell, Daniel C.
• Application Number:
  US16/176,244
• Publication Number:
  US 20200133521A1
• Time in Patent Office: 804 Days
• CPC Class Codes: G06F 2009/45579 I/O management, e.g. providing access to device drivers or storage
• CPC Class Codes: G06F 2009/45583 Memory management, e.g. access or allocation
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0622 in relation to access
• CPC Class Codes: G06F 3/0662 Virtualisation aspects
• CPC Class Codes: G06F 3/0664 at device level, e.g. emulation of a storage device or system
• 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 3/0673 Single storage device
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects

Abstract

Data storage systems of a cluster associate virtual volumes (VVols) with protocol endpoints (PEs) organized into host-specific initiator groups. From the perspective of the hosts, the PEs are specialized SCSI LUNs that aggregate the VVols bound thereto. In operation, a host directs VVol access requests to the host-specific PEs, and a data storage system processes the VVol requests as received via the corresponding PEs.

Claims

1. A method of operating a data storage system in a cluster of storage systems to provide virtual-volume data storage to a plurality of virtual-computing (VC) hosts, the data storage system including first and second processing nodes paired in an active-active manner to provide for (a) shared processing of a workload in a non-failure operating condition, and (b) single-node processing of the workload in a failover operating condition, the method comprising:

organizing physical storage as a plurality of virtual volumes (VVols) each being a virtualized unit of storage for a corresponding virtual machine hosted by a respective VC host;

creating protocol endpoints (PEs) and organizing the PEs into host-specific initiator groups (IGs), each PE being a conglomerate storage device to which a respective set of VVols of the plurality of VVols are to be bound for access by a respective VC host, each IG containing a pair of the PEs for a corresponding VC host, one PE of the pair being advertised to the corresponding VC host as optimized on the first processing node and being advertised to the corresponding VC host as non-optimized on the second processing node, the other PE of the pair being advertised to the corresponding VC host as optimized on the second processing node and being advertised to the corresponding VC host as non-optimized on the first processing node;

binding the sets of VVols to the respective PEs, each VVol of a given set being bound to a corresponding one of the pair of PEs of the corresponding host-specific IG; and

subsequently providing data access to the plurality of VVols from each of the given VC hosts via the respective PEs,

wherein the providing of the data access via the respective PEs includes use of two asymmetric logical unit access (ALUA) paths from each VC host to the PEs of the respective IG, a first ALUA path being a primary access path to the optimized PE on the first processing node, a second ALUA path being a secondary access path to the non-optimized PE on the second processing node, the secondary access path being used when the first processing node for the optimized PE has failed.

2. The method of claim 1, wherein each of the PEs advertised as optimized serves as a respective primary access point for its respective bound set of VVols during normal operation, and wherein each of the PEs advertised as non-optimized serves as a failover or other secondary access point for its respective bound set of VVols in the event of node failure or other unavailability of the primary-access PE of the respective IG.

3. The method of claim 1, wherein the data storage systems are distributed as respective storage subsystems among a corresponding set of converged appliances each including (i) a respective compute subsystem for executing applications, (ii) a respective one of the storage subsystems, and (iii) a respective network subsystem configured and operative to provide cluster interconnection among the converged appliances.

4. The method of claim 3, wherein the first processing node of each converged appliance executes a respective hypervisor as well as a set of PEs of the IGs of the storage subsystem, each hypervisor forming a corresponding one of the VC hosts.

5. The method of claim 4, wherein the storage subsystem of each of the converged appliances provides access to those of the plurality of VVols for both (1) the hypervisors executed on the same respective converged appliance, by internal connections between the compute subsystem and storage subsystem thereof, and (2) the hypervisors executed on others of the converged appliances of the cluster, by the cluster interconnection.

6. The method of claim 5, wherein the storage subsystem of one or more of the converged appliances also provides access to those of the plurality of VVols for one or more external, non-converged hosts by a separate external connection.

7. The method of claim 1, wherein one or more of the plurality of VVols is accessible to multiple of the VC hosts, and is aggregated by respective PEs of the multiple VC hosts accordingly.

8. The method of claim 1, wherein the binding of the sets of VVols to the PEs is according to a static load-balancing scheme in which the respective VC hosts are statically divided into two groups and the sets of VVols for each of the groups are always bound through respective PEs on a respective processing node.

9. The method of claim 1, wherein the binding of the sets of VVols to the PEs is according to a dynamic load balancing scheme in which the binding of a each VVol of the sets of VVols is selected based on dynamic criteria.

10. The method of claim 9, wherein the dynamic criteria includes load balancing across processing nodes and selection of a more lightly loaded processing node for each binding.

11. The method of claim 10, further including monitoring node loading during operation and, upon the node loading becoming unbalanced, performing one or more rebind events to cause VVols of the sets of VVols to be re-bound to nodes having less load.

12. A data storage system for use in a cluster of storage systems to provide virtual-volume data storage to a plurality of virtual-computing (VC) hosts, the data storage system including first and second processing nodes paired in an active-active manner to provide for (a) shared processing of a workload in a non-failure operating condition, and (b) single-node processing of the workload in a failover operating condition, the data storage system including storage devices providing physical storage, interface circuitry for connecting the data storage system to the VC hosts, and storage processing circuitry including first and second processing nodes, the first and second processing nodes being configured and operative to execute corresponding computer program instructions to cause the data storage system to perform a method including:

organizing physical storage as a plurality of virtual volumes (VVols) each being a virtualized unit of storage for a corresponding virtual machine hosted by a respective VC host;

creating protocol endpoints (PEs) and organizing the PEs into host-specific initiator groups (IGs), each PE being a conglomerate storage device to which a respective set of VVols of the plurality of VVols are to be bound for access by a respective VC host, each IG containing a pair of the PEs for a corresponding VC host, one PE of the pair being advertised to the corresponding VC host as optimized on the first processing node and being advertised to the corresponding VC host as non-optimized on the second processing node, the other PE of the pair being advertised to the corresponding VC host as optimized on the second processing node and being advertised to the corresponding VC host as non-optimized on the first processing node;

binding the sets of VVols to the respective PEs, each VVol of a given set being bound to a corresponding one of the pair of PEs of the corresponding host-specific IG; and

subsequently providing data access to the plurality of VVols from each of the given VC hosts via the respective PEs,

wherein the providing of the data access via the respective PEs includes use of two asymmetric logical unit access (ALUA) paths from the VC host to the PEs of the respective IG, a first ALUA path being a primary access path to the optimized PE on the first processing node, a second ALUA path being a secondary access path to the non-optimized PE on the second processing node, the secondary access path being used when the first processing node for the optimized PE has failed.

13. The data storage system of claim 12, wherein each of the PEs advertised as optimized serves as a respective primary access point for its respective bound set of VVols during normal operation, and wherein each of the PEs advertised as non-optimized serves as a failover or other secondary access point for its respective bound set of VVols in the event of node failure or other unavailability of the prim ary-access PE of the respective IG.

14. The data storage system of claim 12, wherein the data storage system is one of a set of data storage systems distributed as respective storage subsystems among a corresponding set of converged appliances each including (i) a respective compute subsystem for executing applications, (ii) a respective one of the storage subsystems, and a respective network subsystem configured and operative to provide cluster interconnection among the converged appliances.

15. The data storage system of claim 14, wherein the first processing node of each converged appliance executes a respective hypervisor as well as a set of PEs of the IGs of the storage subsystem, each hypervisor forming a corresponding one of the VC hosts.

16. The data storage system of claim 15, wherein the storage subsystem of each of the converged appliances provides access to those of the plurality of VVols for both (1) the hypervisors executed on the same respective converged appliance, by internal connections between the compute subsystem and storage subsystem thereof, and (2) the hypervisors executed on others of the converged appliances of the cluster, by the cluster interconnection.

17. The data storage system of claim 16, wherein the storage subsystem of one or more of the converged appliances also provides access to those of the plurality of VVols for one or more external, non-converged hosts by a separate external connection.

18. The data storage system of claim 12, wherein one or more of the plurality of VVols is accessible to multiple of the VC hosts, and is aggregated by respective PEs of the multiple VC hosts accordingly.

19. The data storage system of claim 12, wherein the binding of the sets of VVols to the PEs is according to a static load-balancing scheme in which the respective VC hosts are statically divided into two groups and the sets of VVols for each of the groups are always bound through respective PEs on a respective processing node.

20. The data storage system of claim 12, wherein the binding of the sets of VVols to the PEs is according to a dynamic load balancing scheme in which the binding of each VVol of the sets of VVols is selected based on dynamic criteria.

21. The data storage system of claim 20, wherein the dynamic criteria includes load balancing across processing nodes and selection of a more lightly loaded processing node for each binding.