My mind is still stuck temporarily back in 2002, musing on ONTAP 6.3.
In that release, NetApp introduced SAN support, and we conducted several usability studies around SAN use cases. These studies were structured on the ONTAP CLI, FilerView and SnapDrive.
I remember coming away from those usability studies with a few observations which I'll share.
SAN is not simple
I mean "not simple" in a relative sense. Setting up and managing a fibre channel environment is more complicated than an equivalent NAS solution.
In addition to worrying about all the usual IP-based technologies, SAN administrators must concern themselves with host software and hardware, additional switches, port mappings, and so on.
One of the interesting corollaries is that, even if SAN storage management is simplified, it does not necessarily make SAN simple. You are still at the mercy of all the other details (switches, hosts, HBAs, ...).
Hence short of controlling all the hardware and software end-to-end, it is difficult for a vendor to deliver a holistically simplified SAN solution.
NetApp's SAN solution
Our usability test subjects included customers of other vendors' hardware, as well as existing NetApp customers and SAN novices. Naturally our SAN usability tasks included unstructured tasks around the creation of LUNs.
The test subjects who had the hardest time with the NetApp solution were the experienced SAN administrators. Their prior experience tended to predispose them to seek more complicated work flows.
They spent time poking around trying to figure out how to cobble together a LUN from multiple disks, not initially realizing that NetApp provides an abstraction layer allowing volumes and aggregates to be carved into multiple LUNs.
Once they realized this, the existing SAN administrators were pleased with the relative simplicity of LUN creation on NetApp storage systems.
(The abstraction layer also makes possible products like our storage virtualization V-Series and host consistent SnapShots using SnapDrive).
Cut & Paste matters, too
Two of the most persistently annoying usability issues had little to do with storage and everything to do with multiple vendors and user interface design.
One involved terminology - an intermingling of port names, node names, WWNs, WWPNs, WWNNs, initiators. The labels were not always consistently applied, and varied between vendors (this is an issue that a single vendor can't fix).
The other involved cut and paste issues. A node might be presented as
00A0B301-03DE0000
in one user interface, but another vendor's user interface might insist that it be formatted as
00:A0:B3:01:03:DE:00:00
requiring an infuriating diversion to Notepad for a simple cut/paste operation.
This leads to a simple user interface directive:
Format consistently for presentation, but be permissive for input rules
The more things change ...
In my most recent engineering project, I've been re-examining some SAN work flows and have become reacquainted with each of these issues.
I thought they'd be interesting to share.