Fiber Channel Round Robin Load Balancing

Fiber channel round robin load balancing (FCRR) is a method of utilizing all fiber channel paths available from a given host to a given logical unit number (LUN). In a typical fiber channel environment VI3 hosts will see multiple (generally eight) paths to each LUN. The eight paths travel over two separate host bus adapters (HBAs) which are physically cabled into two separate SAN fabrics. At the storage array, four host ports are cabled into one fabric and four host ports are cabled into the other fabric.
In previous versions any balancing across the available eight paths to a given LUN had to be done manually and often at the command line with a scheduled script.

The vast majority of VMware environments were not implementing any type of fiber channel balancing.

With FCRR, VMware administrators have several options for load balancing their fiber channel activity.
To configure FCCR:
1. When to switch – select the parameters, either max blocks, max commands or both.
2. Which target to use – preferred target, most recently used (MRU) or any available.
3. Which HBA to use – preferred HBA, MRU HBA, HBA with least outstanding I/O requests or any available HBA.
A script will be written to perform the initial configuration of the FCRR, which should not need to be re-configured after the initial script execution. The configuration will be applied to each presented LUN for each VI3 host.

esxcfg-mpath – lun <*.lun> --policy custom –H minq –T any –C 0 –B 2048
The policy states that the LUN should utilize a custom policy that determines which (of two) HBA to utilize based on the minimum queue length. This HBA selection is triggered every 2048 blocks transmitted to a given LUN over the same target. The policy will use any targets available to either of the two HBAs. In using storage that manages host port load-balancing, LUNs will only have two paths (one per fabric) and the storage array will perform storage array host port balancing from within its own management. With other storage arrays it is typical to perform this host port balancing from the hosts accessing the given storage array.
It is also important to note that path or HBA switching does not occur while there are outstanding SCSI commands or a path failover in progress. In these situations the target selection process would commence once the SCSI commands or path failover was complete.