High Availability Virtual Machine Cluster

• US 20080189468A1
• Filed: 01/21/2008
• Published: 08/07/2008
• Est. Priority Date: 02/02/2007
• Status: Active Grant

• Current Assignee: Vmware LLC (Broadcom, Inc.)
• Original Assignee: VMware, Inc. (Broadcom, Inc.)
• Inventors: Schmidt, Rene W., Rajagopal, Sridhar
• Granted Patent: US 8,554,981 B2
• Time in Patent Office: Days
• US Class Current: 711/6
• CPC Class Codes: G06F 11/1084 Degraded mode, e.g. caused by single or multiple storage removals or disk failures
• CPC Class Codes: G06F 11/203 using migration
• CPC Class Codes: G06F 11/2033 switching over of hardware resources
• CPC Class Codes: G06F 11/2035 without idle spare hardware
• CPC Class Codes: G06F 11/2046 where the redundant components share persistent storage G06F11/2043 takes precedence
• CPC Class Codes: G06F 12/08 in hierarchically structured memory systems, e.g. virtual memory systems
• CPC Class Codes: G06F 2009/45587 Isolation or security of virtual machine instances
• CPC Class Codes: G06F 9/455 Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects

Abstract

One embodiment of the present invention is a system including: (a) plural virtualization systems configured in a cluster; (b) storage accessible to each virtualization system of the cluster, wherein for each virtual machine operative in a virtualization system of the cluster, the storage maintains a representation of virtual machine state that includes at least a description of a hardware system virtualized and an image of virtualized memory state for the virtual machine; and (c) a failover system that, responsive to an interruption of, or on, a particular one of the virtualization systems, transitions at least one affected virtual machine to another virtualization system of the cluster and resumes computations of the transitioned virtual machine based on state encoded by a corresponding one of the virtual machine states represented in the storage.

Claims

1. A system comprising:

plural virtualization systems configured in a cluster;

storage accessible to each virtualization system of the cluster, wherein for each virtual machine operative in a virtualization system of the cluster, the storage maintains a representation of virtual machine state that includes at least a description of a hardware system virtualized and an image of virtualized memory state for the virtual machine; and

a failover system that, responsive to an interruption of, or on, a particular one of the virtualization systems, transitions at least one affected virtual machine to another virtualization system of the cluster and resumes computations of the transitioned virtual machine based on state encoded by a corresponding one of the virtual machine states represented in the storage.

2. The system of claim 1, wherein the transitioning by the failover system includes:

starting a new instance of the affected virtual machine using the virtualized hardware system description; and

at least partially defining state of the new instance using the image of virtualized memory state.

3. The system of claim 1,

wherein the image of virtualized memory state for a particular virtual machine is encoded in a volume of the storage that is distinct from, and not commingled with, storage used to encode an image of virtualized memory state for any other virtual machine.

4. The system of claim 1, further comprising:

encapsulations, in respective volumes of the storage, of virtual machine state for respective ones of the virtual machines, wherein each of the encapsulations encodes at least a description of the hardware system virtualized and backing state data its corresponding virtual machine.

5. The system of claim 4,

wherein the respective encapsulations are distinct from, and are not commingled with, each other, and wherein each of the respective volumes is configured for exclusive block level access in connection with operation of its corresponding virtual machine.

6. The system of claim 4,

wherein the respective encapsulations further encode state of one or more virtualized storage resources employed by the corresponding virtual machine.

7. The system of claim 4,

wherein the respective encapsulations further encode one or more snapshots of virtual machine state in addition to a current state.

8. The system of claim 1, wherein the failover system includes:

an availability agent associated with each virtualization system of the cluster, wherein respective availability agents maintain identifiers for respective representations of virtual machine state for at least those ones of the virtual machines which the associated virtualization system may be called upon to continue computations.

9. The system of claim 8,

wherein the availability agents together maintain a coherent replicated representation of identifiers for the encapsulated representations of virtual machine state.

10. The system of claim 8,

wherein the availability agents each maintain, for at least those virtual machines executing on another virtualization system of the cluster, identifiers for the respective encapsulated representations of virtual machine state in the storage.

11. The system of claim 8,

wherein, for each virtual machine executing on one of the virtualization systems of the cluster, one of the availability agents pre-identifies its corresponding virtualization system as the primary failover host.

12. The system of claim 8, further comprising:

a provisioning mechanism that tracks, for each virtual machine, a set of resource requirements, the provisioning mechanism admitting virtual machine computations to the cluster based on an admission control policy that, consonant with the resource requirements of virtual machines provisioned to any particular set of virtualization systems, assures sufficient excess capacity on other virtualization systems of the cluster to support, in the event of interruption of the particular set, transition of all virtual machines operative thereon to the other virtualization systems of the cluster.

13. The system of claim 12,

wherein the resource requirements include both computational and memory requirements.

14. The system of claim 12,

wherein the admission control policy conservatively estimates, for each resource requirement, a worst case capacity sufficient to support the most resource intense virtual machine operative in the cluster, and based on the estimated worst case capacity requirements admits new virtual machine computations only if sufficient excess capacity exists on the virtualization systems of the cluster to support, in the event of interruption of the particular set of virtualization systems, all virtual machines operative thereon.

15. The system of claim 12, wherein the particular set is selected based on one or more of:

number of interruptions or failures to be tolerated; and

specialized resources and/or connectivity of any particular virtualization system of the cluster.

16. The system of claim 12,

wherein the particular set numbers at least one;

and

wherein the particular set includes at least the most heavily provisioned virtualization system.

17. The system of claim 12,

wherein the particular set is sufficient to span multiple virtualization system failures.

18. The system of claim 1, further comprising:

a distributed resource scheduling mechanism that assigns virtual machines to virtualization systems based on actual runtime load conditions and precomputes, for each virtual machine, a preferred alternative virtualization system based on the actual runtime load conditions.

19. The system of claim 1,

wherein the storage includes a storage area network (SAN);

and

wherein the storage unit is provisioned and maintained as a storage unit of the SAN.

20. The system of claim 1,

wherein the storage is configured as network attached storage (NAS) or is interfaced using a small computer systems interface over TCP/IP (iSCSI) protocol.

21. The system of claim 1,

wherein the virtualization systems each execute on an underlying hardware platform;

and

wherein the interruption includes one or more of;

a failure of the underlying hardware platform or connectivity of the underlying hardware platform with a network, device or resource;

a failure of the particular virtualization system, and

a failure of the virtual machine to be transitioned.

22. The system of claim 1,

wherein the interruption includes scheduled or unscheduled power down of the virtual machine to be transitioned, of the particular virtualization system or of an underlying hardware platform.

23. A high-availability cluster of virtualization systems in which failover is supported at least in part by maintaining for each virtual machine an encapsulated representation of virtual machine state, including at least a description of the hardware system virtualized together with backing state therefor, in storage accessible to each virtualization system of the cluster, and by re-instantiating in response to a virtualization system failure and based on the respective descriptions of hardware systems virtualized, one or more virtual machines and resuming respective computations thereof based on the respective backing states.

24. The high-availability cluster of claim 23, further comprising:

a cluster management system that tracks, for each virtual machine, a set of resource requirements, the cluster management system admitting virtual machine computations to the cluster based on an admission control policy that, consonant with the resource requirements of virtual machines provisioned to any particular set of the virtualization systems, assures sufficient excess capacity on other virtualization systems of the cluster to support, in the event of failure of the particular set, the re-instantiation and resumption of all virtual machines operative thereon on the other virtualization systems of the cluster.

25. The high-availability cluster of claim 23, further comprising:

a cluster management system that selects failover targets for the virtual machines of the cluster and propagates the failover target selections to respective failover agents executing in conjunction with each of the virtualization systems of the cluster.

26. A method of operating a cluster of virtualization systems, the method comprising:

maintaining, for each virtual machine executing on a virtualization system of the cluster, an encapsulated representation of virtual machine state in storage accessible to each virtualization system of the cluster, wherein the virtual machine state includes at least a description of a hardware system virtualized and an image of virtualized memory state for the corresponding virtual machine; and

transitioning at least one of the virtual machines to another virtualization system of the cluster and resuming computations of the transitioned virtual machine based on state encoded by a corresponding one of the virtual machine states represented in the storage.

27. The method of claim 26,

wherein the transitioning is in response to an interruption of, or on, a particular one of the virtualization systems.

28. The method of claim 26,

wherein the encapsulated representations are distinct from, and are not commingled with, each other, and

further comprising configuring respective ones of the encapsulated representations for exclusive block level access in connection with operation of a corresponding one of the virtual machines.

29. A computer program product encoded in one or more computer readable media, the product comprising:

an availability agent executable in association with respective virtualization systems of a cluster, the availability agent including an instruction sequence executable to resume, on the associated virtualization system, execution of a virtual machine that previously executed on another virtualization system of the cluster,

wherein the availability agent maintains, for each virtual machine for which the associated virtualization system may be called upon to resume computations, an identifier for identifying in shared storage a representation of virtual machine state that includes at least a description of the a hardware system virtualized and an image of virtualized memory state.

30. The computer program product of claim 29, further comprising:

an instruction sequence executable in association with a particular virtualization system to detect a failure of, or on, another one of the virtualization systems and, responsive thereto, to initiate the resumption.

31. The computer program product of claim 29, wherein the identifiers maintained by availability agent identifies a corresponding encapsulated representation of virtual machine state and allow a failover target to assume exclusive block level access to the virtual machine state in connection with the resumption of computations for a corresponding one of the virtual machines.

32. The computer program product of claim 29, wherein the computer readable media are selected from the set of a disk, tape or other magnetic, optical or electronic storage medium.

33. The computer program product of claim 29, at least transiently encoded in the computer readable media in connection with transmission via network, wire line, wireless or other communications.