Load balancing in a virtualized computing environment based on a fabric limit

• US 9,973,433 B2
• Filed: 08/28/2015
• Issued: 05/15/2018
• Est. Priority Date: 07/10/2015
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Matzek, Samuel D., Tenner, Jeffrey W., McBrearty, Gerald F.
• Primary Examiner(s): Gillis, Brian J
• Assistant Examiner(s): Lin, Steve
• Application Number:
  US14/839,651
• Publication Number:
  US 20170012896A1
• Time in Patent Office: 991 Days
• CPC Class Codes: H04L 41/5041 characterised by the time relationship between creation and deployment of a service
• CPC Class Codes: H04L 47/125 by balancing the load, e.g. traffic engineering
• CPC Class Codes: H04L 49/505 Corrective measures
• CPC Class Codes: H04L 49/70 Virtual switches
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• CPC Class Codes: H04L 67/1097 for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]

Abstract

A method manages virtual machines in a virtualized computing environment based at least in part on limitations associated with storage fabrics through which virtual machines may access one or more storage systems in such an environment. The storage fabric limitations may be used, for example, in connection with placing virtual machines on hosts in a virtualized computing environment. As another example, storage fabric limitations may be used in connection with deploying virtual machines into a virtualized computing environment to reduce the likelihood of boot errors. As still another example, storage fabric limitations may be used in connection with load balancing across multiple fabrics in a virtualized computing environment.

Claims

1. A method of load balancing in a computing environment of the type including a plurality of fabrics coupling a plurality of hosts to one or more storage systems, each fabric among the plurality of fabrics coupling one or more of the plurality of hosts to one or more of the one or more storage systems, the method comprising:

determining a fabric limit for a limited fabric resource of each of the plurality of fabrics, wherein the limited fabric resource for each respective fabric among the plurality of fabrics relates to a relative load and/or capacity of the respective fabric, and wherein the fabric limit is an active login limit, an active discovery limit, a zone limit, or a host limit;

determining an overcommitted fabric capacity for each of the plurality of fabrics from the fabric limits;

determining an allocation of the limited fabric resource for each of the plurality of fabrics; and

balancing utilization of the limited fabric resource across the plurality of fabrics using at least one processor based at least in part upon the determined fabric limits and determined allocations of the limited fabric resource for the plurality of fabrics, wherein balancing utilization of the limited fabric resource includes migrating at least one virtual machine that is resident on a first host among the plurality of hosts and that utilizes a first fabric among the plurality of fabrics to use a second fabric among the plurality of fabrics, wherein the at least one virtual machine that is migrated to use the second fabric remains resident on the first host after migration, wherein balancing utilization of the limited fabric resource across the plurality of fabrics is based at least in part on the determined overcommitted fabric capacity for each of the plurality of fabrics, and wherein migrating the at least one virtual machine includes, for a first virtual machine;

powering down the first virtual machine;

reconfiguring the first virtual machine to utilize the second fabric instead of the first fabric; and

powering up the first virtual machine such that the first virtual machine utilizes the second fabric instead of the first fabric.

2. The method of claim 1, wherein the fabric includes a Fibre Channel switch, and wherein the fabric limit is an active Fabric Login (FLOGI) limit or an active Fabric Discovery (FDISC) limit.

3. The method of claim 1, wherein the limited fabric resource comprises a fabric login for a storage switch in the first fabric, and wherein the fabric limit comprises an active login limit of the storage switch.

4. The method of claim 3, wherein determining the fabric limit comprises querying the storage switch to retrieve the fabric limit.

5. The method of claim 3, wherein determining the allocation of the limited fabric resource comprises accessing the storage switch to retrieve a current allocation of fabric logins for the storage switch.

6. The method of claim 3, wherein determining the allocation of the limited fabric resource comprises tracking active fabric logins in the computing environment.

7. The method of claim 1, wherein determining the overcommitted fabric capacity includes adjusting the fabric limit by a fabric overcommit ratio.

8. The method of claim 1, wherein balancing utilization of the limited fabric resource across the plurality of fabrics further comprises migrating at least one virtual machine to an alternate storage access technology to free up fabric resources on at least one of the plurality of fabrics.

9. The method of claim 8, wherein migrating the at least one virtual machine to the alternate storage access technology is based at least in part on the at least one virtual machine having a low priority.

10. The method of claim 8, wherein the fabric comprises a Fibre Channel-based network, and wherein the alternate storage access technology comprises vSCSI.

11. The method of claim 1, wherein determining the fabric limit, determining the allocation of the limited fabric resource, and balancing utilization of the limited fabric resource across the plurality of fabrics are performed periodically.

12. The method of claim 1, wherein determining the fabric limit, determining the allocation of the limited fabric resource, and balancing utilization of the limited fabric resource across the plurality of fabrics are performed in response to a virtual machine deployment operation performed when at least one fabric lacks capacity.