Cloud defragmentation

• US 9,207,961 B2
• Filed: 03/13/2013
• Issued: 12/08/2015
• Est. Priority Date: 05/25/2012
• Status: Active Grant

• Current Assignee: Rackspace US Incorporated (Apollo Global Management, Inc.)
• Original Assignee: Rackspace US Incorporated (Apollo Global Management, Inc.)
• Inventors: Conrad, Jeffrey Keating, Kirkland, Gregory Scott
• Primary Examiner(s): Kim, Sisley
• Application Number:
  US13/798,568
• Publication Number:
  US 20130318526A1
• Time in Patent Office: 1,000 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06F 2009/4557 Distribution of virtual machine instances; Migration and load balancing
• CPC Class Codes: G06F 9/45533 Hypervisors; Virtual machine monitors
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects

Abstract

In an embodiment, a method includes identifying, in a defragmentation scheduling logic of a first system of a data center, a first virtual machine (VM) of a first server of the data center to select for migration based on a cost determined according to a source algorithm, identifying a second server of the data center to select for receipt of the first VM based on a cost determined according to a destination algorithm, where the second server has available free space to receive the first VM and the available free space has been scrubbed prior to the identification, and migrating a substantial plurality of VMs between servers of the data center based on cost according to the source and destination algorithms, to defragment the data center.

Claims

1. A method comprising:

identifying, in a defragmentation scheduling logic of a first system of a data center, a first virtual machine (VM) of a first server of the data center to select for migration based on a cost determined according to a source algorithm;

identifying a second server of the data center to select for receipt of the first VM based on a cost determined according to a destination algorithm, wherein the source algorithm and the destination algorithm are interdependent on each other, and the second server has available free space to receive the first VM and the available free space has been scrubbed prior to the identification of the second server;

migrating the first VM from the first server to the second server;

performing a scrubbing operation on the first server after migrating the first VM;

associating the first VM with a second type of hypervisor in the second server, wherein the first VM is associated with a first type of hypervisor in the first server, the first hypervisor type heterogeneous to the second hypervisor type, wherein the second hypervisor type is an update to the first hypervisor type;

responsive to a determination by the destination algorithm that a large contiguous space is needed responsive to presence of one or more VMs of the first server to be migrated, the one or more VMs of a size that exceeds a threshold, migrating additional VMs from the first server to another server;

migrating a substantial plurality of VMs between servers of the data center based on cost according to the source and destination algorithms, to defragment the data center to enable greater amounts of contiguous available space for VMs subsequent to the data center defragmentation; and

maintaining computing resources of the data center including the substantial plurality of VMs accessible to customers of the data center during the migrating.

2. The method of claim 1, wherein performing the scrubbing operation removes data fragments of a prior VM that executed on the second first server.

3. The method of claim 1, further comprising performing a scrubbing operation on all potential destination servers before identifying the second server.

4. The method of claim 1, further comprising:

iteratively migrating each of a plurality of VMs from the first server to other servers of the data center and scrubbing the first server; and

thereafter making the first server available for receipt of migration by the destination algorithm.

5. The method of claim 1, wherein the method comprises migrating more than 70,000 VMs within several weeks.

6. An article comprising a non-transitory computer-readable storage medium comprising instructions to:

identify, in a defragmentation scheduling logic of a first system of a data center, a first virtual machine (VM) of a first server of the data center to select for migration based on a cost determined according to a source algorithm;

identify a second server of the data center to select for receipt of the first VM based on a cost determined according to a destination algorithm, wherein the source algorithm and the destination algorithm are interdependent on each other, and the second server has available free space to receive the first VM and the available free space has been scrubbed prior to the identification of the second server;

migrate the first VM from the first server to the second server;

perform a scrubbing operation on the first server after migration of the first VM;

associate the first VM with a second type of hypervisor in the second server, wherein the first VM is associated with a first type of hypervisor in the first server, the first hypervisor type heterogeneous to the second hypervisor type, wherein the second hypervisor type is an update to the first hypervisor type;

responsive to a determination by the destination algorithm that a large contiguous space is needed responsive to presence of one or more VMs of the first server to be migrated, the one or more VMs of a size that exceeds a threshold, migrate additional VMs from the first server to another server; and

migrate a substantial plurality of VMs between servers of the data center based on cost according to the source and destination algorithms, to defragment the data center to enable greater amounts of contiguous available space for VMs subsequent to the data center defragmentation, wherein computing resources of the data center including the substantial plurality of VMs are accessible to customers of the data center during the migrating.

7. The article of claim 6, wherein the scrubbing operation removes data fragments of a prior VM that executed on the second first server.

8. A data center comprising:

a plurality of servers each to execute a hypervisor on which one or more virtual machines (VMs) operate; and

a data center management system configured to;

identify a first VM of a first server to select for migration based on a cost determined according to a source algorithm;

identify a second server to select for receipt of the first VM based on a cost determined according to a destination algorithm, wherein the source algorithm and the destination algorithm are interdependent on each other and the second server has available free space to receive the first VM and the available free space has been scrubbed prior to the identification of the second server;

migrate the first VM from the first server to the second server;

perform a scrubbing operation on the first server after migration of the first VM;

associate the first VM with a second type of hypervisor in the second server, wherein the first VM is associated with a first type of hypervisor in the first server, the first hypervisor type heterogeneous to the second hypervisor type, wherein the second hypervisor type is an update to the first hypervisor type;

responsive to a determination by the destination algorithm that a large contiguous space is needed responsive to presence of one or more VMs of the first server to be migrated, the one or more VMs of a size that exceeds a threshold, migrate additional VMs from the first server to another server; and

migrate a substantial plurality of VMs between servers of the data center based on cost according to the source and destination algorithms, to defragment the data center to enable greater amounts of contiguous available space for VMs subsequent to the data center defragmentation.

9. The data center of claim 8, wherein the substantial plurality of VMs are accessible to customers of the data center during the migration.