REPLICATION OF ASSETS ACROSS DATA CENTERS

• US 20140207736A1
• Filed: 01/18/2013
• Published: 07/24/2014
• Est. Priority Date: 01/18/2013
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Susanto, Ferry, Dournov, Pavel, Wright, Eron D., Ghani, Usman, Rajaram, Koushik
• Granted Patent: US 9,298,790 B2
• Time in Patent Office: Days
• US Class Current: 707/634
• CPC Class Codes: G06F 16/27 Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
• CPC Class Codes: G06F 16/273 Asynchronous replication or reconciliation
• CPC Class Codes: G06F 16/275 Synchronous replication
• CPC Class Codes: H04L 67/1095 Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes

Abstract

The replication of an asset from a source cluster in a source data center to multiple target clusters in multiple destination data centers. The replication occurs by first estimating or determining a cost parameter associated with copying of the asset from the source cluster to each of at least some of the target clusters. As an example, the cost parameter might be a geographical parameter, but might also be any cost parameter such as a channel bandwidth, channel cost, utilization ratio or the like. Based on the cost parameters, an order of replication priority is determined. Then, the replication is initiated in accordance with the prioritization. The replication may occur in multiple phases, and replication prioritization may occur on a per phase basis.

Claims

1. A method for replicating an asset from a source cluster in a source data center to a plurality of target clusters in a plurality of destination data centers, the method comprising:

an act of at least estimating a cost parameter associated with copying of the asset from the source cluster to each of at least some of the target clusters;

an act of prioritizing an order of replication based on the cost parameters; and

an act of initiating the replication in accordance with the prioritization.

2. The method in accordance with claim 1, wherein the cost parameter for at least one of the target clusters comprises a geographical relationship between the source cluster and the target cluster.

3. The method in accordance with claim 1, wherein the replication occurs in a plurality of phases.

4. The method in accordance with claim 3, wherein a number of copy operations at least of a first type are limited to a first limit per replication phase.

5. The method in accordance with claim 4, wherein a number of copy operations of a second type are limited to a second limit per replication phase.

6. The method in accordance with claim 4, wherein the number of copy operations of a second type are not limited per replication phase.

7. The method in accordance with claim 3, wherein the prioritization of the order of replication occurs one replication phase at a time.

8. The method in accordance with claim 3, wherein the cost parameter for at least one of the target clusters comprises a geographical relationship between the source cluster and the target cluster, wherein at least one remotely located target cluster within a given remote region is prioritized for copying of the asset within an initial replication phase.

9. The method in accordance with claim 8, wherein the at least one remotely located target cluster is located on a different continent than the source cluster.

10. The method in accordance with claim 1, wherein the cost parameter for at least one of the target clusters comprises a bandwidth of the channel between the source cluster and the target cluster.

11. The method in accordance with claim 1, wherein the cost parameter for at least one of the target clusters comprises a utilization of the channel between the source cluster and the target cluster.

12. The method in accordance with claim 1, wherein the asset is a virtual machine image.

13. The method in accordance with claim 1, wherein the asset is an application package.

14. The method in accordance with claim 1, further comprising:

an act of tracking a state of replication associated with each of the at least some of the target clusters.

15. The method in accordance with claim 1, further comprising:

an act of receiving the asset into the source cluster.

16. The method in accordance with claim 15, further comprising:

an act of receiving a command to replicate, wherein the act of at least estimating and the act of prioritizing occurs in response to the command to replicate.

17. The method in accordance with claim 1, further comprising:

an act of locking each replicated copy of the asset after replication so that writing to the replicated copy of the asset or deleting the replicated asset is prohibited.

18. The method in accordance with claim 17, wherein if a lock of a replicated asset on a particular target cluster is broken, the replicated asset in the particular target cluster is deleted, and marked for re-replication from another cluster that has a copy of the asset.

19. A computer program product comprising one or more computer-readable storage media having thereon computer-executable instructions that, when executed by one or more processors of a computing system, cause the computing system to perform a method for replicating an asset from a source cluster in a source data center to a plurality of target clusters in a plurality of destination data centers, the method comprising:

an act of at least estimating a cost parameter associated with copying of the asset from the source cluster to each of at least some of the target clusters;

an act of prioritizing an order of replication based on the cost parameters; and

an act of initiating the replication in accordance with the prioritization.

20. A system comprising a replication engine configured to perform replication of an asset from a source cluster in a source data center to a plurality of target clusters in a plurality of destination data centers by performing the following:

an act of at least estimating a cost parameter associated with copying of the asset from the source cluster to each of at least some of the target clusters;

an act of prioritizing an order of replication based on the cost parameters; and

an act of initiating the replication in accordance with the prioritization.