PowerStore: "Appliance port pairs are connected to the same data switch" error when multiple independent iSCSI switches are used

The PowerStore OS may generate alerts when multiple independent iSCSI switches are used. 

The PowerStore Operating System (OS) has a subsystem called ONV (Ongoing Network Validation).  ONV is designed to identify configurations which may not be optimal. Some valid configurations may generate false positive warnings.

The PowerStore storage networks generally involve pairs of Ethernet switches for iSCSI connectivity with some form of inter-switch link (ISL) between them. In this configuration, Storage networks typically span both switches through the ISL. Corresponding ports from each PowerStore node connect to a different switch. For example, I/O Card 1 port 2 on node A and I/O Card 1 port 2 on node B (known collectively as a ‘Port Pair’) go to two different switches. This gives each port pair redundancy for switch/cable failure.

Figure 1: Storage networks with connected switches

 

Some customers prefer to have independent iSCSI switches (with no ISLs) for their storage networks. Without ISLs, VLANs cannot span the two switches – each switch runs its own subnet. The PowerStore assigns storage networks by port pair across nodes, so this method requires the same port from each node to be connected to the same switch. The first switch would be connected to one port pair, and the second switch connects to a different port pair on the appliance. When cabled in this manner the ONV detects that a single port pair has no redundancy at a switch level and generates a warning, even though the overall configuration is fully redundant.
 

Figure 2: Storage networks with independent switches

The warnings generated by the PowerStore ONV subsystem do not necessarily mean there is anything wrong with the configuration, nor does it mean this configuration does not follow best practices.

As long as there are two switches with two storage networks connected, and each host has a connection to both storage networks, the cluster can operate without interruption through a single failure (such as a switch). In this configuration, the warnings may be safely ignored. When these warnings are ignored, care should be taken to ensure the configuration really is fully redundant.