HP Matrix Operating Environment and HP Storage Provisioning Manager support for port groups

Technical white paper | Matrix OE and SPM

storage on the storage system and administrators overseeing the storage system itself. The steps to be performed are

indicated here, without distinction regarding who should perform which step.

1. Identify controller ports in need of load balancing (using storage system management tools)

2. Identify servers (initiator WWNs) that need to be moved to new controller ports

a. Identify list of server WWNs that need rezoning to new controller ports (use storage management console to

build the initiator list)

b. Determine which Matrix services/logical servers ‘own’ these initiator WWNs and if the volumes accessed via

those initiators are redundant or not (use the Matrix OE visualization logical server summary page)

3. Based on above information, determine pre-zoning actions. This includes identifying if servers need to be taken

down, and if VMs need to be moved to new hosts.

4. Perform the rezoning to the new controller ports (Use the provided PowerShell script to perform the rezoning

operations.)

5. Final actions (This includes starting servers back up and moving VMs back that may have been moved.)

The steps are covered in further detail below.

Identify controller ports in need of load balancing

Determining if load balancing is needed is beyond the scope of this paper. However, there are multiple criteria that can be

used to determine which controller ports need offloading. It can be based on the number of hosts currently connected to

each controller port, where the goal is to keep this number level across all controller ports. Another option is to use actual

I/O traffic through each controller port. Often, the storage solution will provide tools to monitor this information. The

screenshot in Figure 2 shows the monitoring of disk I/O traffic across the controller ports for HP 3PAR StoreServ.

Figure 2. HP 3PAR StoreServ Management Console showing total throughput over time for each controller port.

Regardless of the criteria used, two things need to be identified:

1. What controller ports are overloaded and need servers moved off of them

2. What controller ports are underutilized and can have servers migrated to them