VDI Storage overcommit and rebalancing

• US 8,307,187 B2
• Filed: 09/11/2009
• Issued: 11/06/2012
• Est. Priority Date: 09/12/2008
• Status: Active Grant

• Current Assignee: Vmware LLC (Broadcom, Inc.)
• Original Assignee: VMware, Inc. (Broadcom, Inc.)
• Inventors: Jin, Ke, Taylor, Frank, Johnston, Keith, Patel, Amit, Chawla, Puneet Singh
• Primary Examiner(s): Dudek, Jr., Edward
• Assistant Examiner(s): Rossiter, Sean D
• Application Number:
  US12/558,443
• Publication Number:
  US 20100070978A1
• Time in Patent Office: 1,152 Days
• Field of Search: None
• US Class Current: 711/173
• CPC Class Codes: G06F 9/5077 Logical partitioning of resources; Management or configuration of virtualized resources specific details on emulation or internal functioning of virtual machines G06F9/455

Abstract

A method for managing storage for a desktop pool is described. The desktop pool includes a plurality of virtual machines (VMs), each VM having at least one virtual disk represented as a virtual disk image file on one of a plurality of datastores associated with the desktop pool. To identify a target datastore for a VM, a weight of each datastore is calculated. The weight may be a function of a virtual capacity of the datastore and the sum of maximum sizes of all the virtual disk image files on the datastore. The virtual capacity is a product of the data storage capacity of the datastore and an overcommit factor assigned to the datastore. The target datastore is selected as the datastore having the highest weight. The VM may is moved to or created on the target datastore.

Claims

1. A method for managing storage for a desktop pool, the desktop pool comprising a plurality of VMs, each having at least one virtual disk represented as a virtual disk image file on one of a plurality of datastores associated with the desktop pool, the datastores each being a logical storage volume of one or more storage system in electronic communication with a virtual desktop manager server that performs the method, the method comprising:

calculating a weight of each datastore, the weight being a function of a virtual capacity of the datastore and the sum of maximum sizes of virtual disk image files on the datastore, the virtual capacity being a product of the data storage capacity of the datastore and an overcommit factor assigned to the datastore, the overcommit factor being selected by an administrator;

selecting as a target datastore, one datastore of the plurality of datastores, the selecting being based on the weights of the datastores;

placing data files for a VM that is a member of the desktop pool on the target datastore.

2. The method of claim 1, wherein the placing of the data files comprises:

creating the VM by cloning a master VM; and

storing the VM on the target datastore.

3. The method of claim 2, wherein the VM is thin-provisioned such that a virtual disk image file belonging to the VM is smaller than a capacity of a virtual disk represented by the virtual disk image file belonging to the VM.

4. The method of claim 2, wherein the creating of the VM from the master VM comprises creating a delta disk from a master disk image file, the delta disk representing a virtual disk of the VM and defining only differences between the virtual disk of the VM and the master disk;

the target datastore not containing the master disk image file.

5. The method of claim 1, wherein the placing further comprises moving the VM from a source datastore to the target datastore as a part of a datastore rebalancing operation.

6. The method of claim 5, further comprising:

adjusting a weight of the source datastore to reflect a weight the source datastore would be without the VM being present on the source datastore, and performing the selecting based on the adjusted weight of the source datastore.

7. The method of claim 5, wherein the moving of the VM is preformed using hot storage migration when the VM is powered on.

8. The method of claim 1, further comprising:

receiving a datastore assignment from an administrator, the datastore assignment designating the plurality of datastores that are associated with the desktop pool;

for at least one of the datastores, receiving a designation of a disk type restriction, the disk type restriction identifying a type of virtual disk image file is permitted to be placed on the datastore; and

the selecting further comprising identifying, for each type of virtual disk associated with the VM, a datastore that permits the type of virtual disk to be placed on the datastore.

9. The method of claim 8, wherein the VM includes an OS disk and a user data disk, the OS disk and user data disks being types of virtual disks, and the selecting identifies a first datastore designated for OS disks to place a virtual disk image file corresponding to the OS disk types and a second datastore designated for user data disk types to place a virtual disk image file corresponding to the user data disk.

10. The method of claim 1, wherein the weight is calculated as the virtual capacity minus the virtual used capacity at provisioning time minus the virtual used capacity at runtime, the virtual used capacity at provisioning time being the sum of the maximum sizes of the virtual disk image files on the datastore and the virtual used capacity at runtime including disk space required for hypervisor swap files.

11. The method of claim 1, further comprising:

identifying a policy for inhibiting growth of virtual disks by performing one or more operations to reduce the space required of virtual disk image files on a datastore, the policy including a schedule to perform the operations, the operations including one of a disk cleanup and reverting the VM to an original pristine state.

12. A non-transitory machine readable storage medium for managing storage for a desktop pool, the desktop pool comprising a plurality of VMs, each having at least one virtual disk represented as a virtual disk image file on one of a plurality of datastores associated with the desktop pool, the datastores each being a logical storage volume of one or more storage system in electronic communication with a virtual desktop manager server that performs the method, the machine readable storage medium comprising program instructions causing a computer system to implement a method, the method comprising:

calculating a weight of each datastore, the weight being a function of a virtual capacity of the datastore and the sum of maximum sizes of virtual disk image files on the datastore, the virtual capacity being a product of the data storage capacity of the datastore and an overcommit factor assigned to the datastore, the overcommit factor being selected by an administrator;

selecting as a target datastore, one datastore of the plurality of datastores, the selecting being based on the weights of the datastores;

placing data files for a VM that is a member of the desktop pool on the target datastore.

13. The machine readable storage medium of claim 12, wherein the placing of the data files comprises:

creating the VM by cloning a master VM; and

storing the VM on the target datastore.

14. The machine readable storage medium of claim 13, wherein the VM is thin-provisioned such that a virtual disk image file belonging to the VM is smaller than a capacity of a virtual disk represented by the virtual disk image file belonging to the VM.

15. The machine readable storage medium of claim 13, wherein the creating of the VM from the master VM comprises creating a delta disk from a master disk image file, the delta disk representing a virtual disk of the VM and defining only differences between the virtual disk of the VM and the master disk;

and the target datastore does not contain the master disk image file.

16. The machine readable storage medium of claim 12, wherein the placing further comprises moving the VM from a source datastore to the target datastore as a part of a datastore rebalancing operation.

17. The machine readable storage medium of claim 16, wherein the method further comprises:

adjusting a weight of the source datastore to reflect a weight the source datastore would be without the VM being present on the source datastore, and performing the selecting based on the adjusted weight of the source datastore.

18. The machine readable storage medium of claim 16, wherein the moving of the VM is preformed using hot storage migration when the VM is powered on.

19. The machine readable storage medium of claim 12, wherein the method further comprises:

receiving a datastore assignment from an administrator, the datastore assignment designating the plurality of datastores that are associated with the desktop pool;

for at least one of the datastores, receiving a designation of a disk type restriction, the disk type restriction identifying a type of virtual disk image file is permitted to be placed on the datastore; and

the selecting further comprising identifying, for each type of virtual disk associated with the VM, a datastore that permits the type of virtual disk to be placed on the datastore.

20. The machine readable storage medium of claim 19, wherein the VM includes an OS disk and a user data disk, the OS disk and user data disks being types of virtual disks, and the selecting identifies a first datastore designated for OS disks to place a virtual disk image file corresponding to the OS disk types and a second datastore designated for user data disk types to place a virtual disk image file corresponding to the user data disk.

21. The machine readable storage medium of claim 12, wherein the weight is calculated as the virtual capacity minus the virtual used capacity at provisioning time minus the virtual used capacity at runtime, the virtual used capacity at provisioning time being the sum of the maximum sizes of the virtual disk image files on the datastore and the virtual used capacity at runtime including disk space required for hypervisor swap files.

22. The machine readable storage medium of claim 12, the method further comprising:

identifying a policy for inhibiting growth of virtual disks by performing one or more operations to reduce the space required of virtual disk image files on a datastore, the policy including a schedule to perform the operations, the operations including one of a disk cleanup and a VM revert.