Regarding subdividing external LUNs. Our best practice is to not subdivide external storage because it makes it simpler to isolate & monitor flows to the external virtualized storage. And as you point out, this also offers the customer a choice to de-virtualize the external storage if he or she wishes too – preserving future storage options.

It’s a small point, but the figures could show you can support multiple externally attached arrays on a single port just as you can with server side connections.

I see that you have highlighted the problem of having to migrate like-to-like LUNs and thus potentially perpetuating bad configurations. However, there is an easy fix to this problem. Tiered Storage Manager allows you to move online data non-disruptively between unlike LUNs to adjust disk configurations. Likewise with Hitachi Dynamic Provisioning you can reclaim unused storage and increase LUN sizes.

We agree that proper planning is required for any enterprise storage environments especially if it is replicated. But the incremental work for virtualizing an array, although non trivial, is not large compared to the basic work required in any case.

Taken as a whole, I think this write-up is useful for someone contemplating retiring or repurposing old storage arrays. Without a virtualization engine such as the USP doing migration is an exponentially more complex task fraught with both human and IT risks.

Agreed, use cases would make sense. Haven't had a chance to work on this as I've been tied up on other stuff.

The whole thick->thin issue is a great one. I was in a forum before Christmas discussing this exact subject. Lots of interesting comments relating to the subject of TP but lots of nervousness about migrating into a TP environment rather than having a greenfield TP array.

Thx for the feedback, I'll be getting to it soon.