High availability virtual machine cluster

1. A system comprising:

plural virtualization systems configured in a cluster for operations of virtual machines;

storage shared by the virtualization systems of the cluster, wherein the storage maintains encapsulated representations of virtual machine state of the virtual machines, wherein each of the encapsulated representations encodes at least one or more virtual disks for a particular virtual machine of the cluster; and

a failover system that transitions at least one virtual machine affected by an interruption to another virtualization system of the cluster.

2. The system of claim 1 , wherein each of the encapsulated representations is configured to limit guest access to data residing in the one or more virtual disks.

3. The system of claim 1 , wherein one of the encapsulated representations of virtual machine state includes at least a description of a hardware system virtualized and an image of virtualized memory state for a virtual machine, and 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.

4. The system of claim 1 , wherein one of the encapsulated representations of virtual machine state includes at least a description of a hardware system virtualized and an image of virtualized memory state for a virtual machine, and 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.

5. The system of claim 1 , wherein each of the encapsulated representations encodes at least a description of a hardware system virtualized and backing state data of its corresponding virtual machine

6. The system of claim 1 ,

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

7. The system of claim 1 ,

wherein the encapsulated representations further encode state of one or more virtualized storage resources employed by the corresponding virtual machine

8. The system of claim 1 ,

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

9. 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 encapsulated 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.

10. The system of claim 9 ,

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

11. The system of claim 9 ,

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.

12. The system of claim 9 ,

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.

13. The system of claim 9 , 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.

14. The system of claim 13 ,

wherein the resource requirements include both computational and memory requirements.

15. The system of claim 13 ,

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.

16. The system of claim 13 , 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.

17. The system of claim 13 ,

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

18. The system of claim 13 ,

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

19. 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.

20. The system of claim 1 ,

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

21. 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.

22. 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.

23. The system of claim 1 ,

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

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

maintaining, for virtual machines executing on the virtualization systems of the cluster, encapsulated representations of virtual machine state in storage shared by the virtualization systems of the cluster; and

transitioning at least one of the virtual machines to another virtualization system of the cluster,

wherein each of the encapsulated representations encodes one or more virtual disks for a particular virtual machine of the cluster.

25. The method of claim 24 , further comprising configuring each of the

encapsulated representation to limit guest access to data residing in the one or more virtual disks.

26. The method of claim 24 ,

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

27. The method of claim 24 ,

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

28. A non-transitory computer readable storage medium embodying computer software implementing a method of operating a cluster of virtualization systems, the method comprising:

maintaining, for virtual machines executing on the virtualization systems of the cluster, encapsulated representations of virtual machine state in storage shared by the virtualization systems of the cluster; and

transitioning at least one of the virtual machines to another virtualization system of the cluster,

wherein each of the encapsulated representation encodes one or more virtual disks for a particular virtual machine of the cluster.

29. The computer readable storage medium of claim 28 , the method further comprising configuring each of the encapsulated representation to limit guest access to data residing in the one or more virtual disks.

30. The computer readable storage medium of claim 28 ,

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

31. The computer readable storage medium of claim 28 ,

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

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