2009-09-14 02:18 PM
I have a customer who is looking at using a Windows server with an inexpensive iSCSI array to provide cheap/low-end file/archival storage. Given they do already have a FAS system that can handle extra SATA shelves, we've discussed this at length over the last couple weeks.
While I do really like CIFS functionality off NetApp and we have lot of customers using it (for reasons like snapshots (previous versions in particular), snapmirror, volume expansion, quotas (with Operations Manager reporting), no patching, high availability for free given a cluster, deduplication, RAID DP, etc.), I can understand the desire for a dirt cheap lower storage tier and don't want to be dogmatic that "all data must be stored on the NetApp" (although see the prior list for why I think it's a good idea ).
Given that, I wanted to throw this out there to the community -- is anyone else doing this/seen this? If so, have you run into any showstoppers or major issues (especially ones that didn't seem to be of huge consequence at first but were later).
General conversation is quite welcome as well....I just want to be sure I'm giving as complete info as possible in a low-key way in helping them figure out what makes sense.
2009-09-14 03:14 PM
Yeah, OK - so you may look at it from many different angles & get to very different end results. We all know the story around snaps, de-dupe & all this storage heaven
However at the bottom end of the spectrum it is quite often down to personal preferences & existing skill set. So for a 100% Windows guy, just a Windows file server might be a "better" option, than UNIX-ish type of appliance which "accidentally" can do CIFS as well.
In your particular scenario though I can see one, fairly obvious, benefit of NetApp - it would be a single point of management vs. a Windows server plus an iSCSI array. That would be not so obvious if the customer is thinking about something even simpler, i.e. a server with internal drives.
2009-09-15 11:59 AM
What this customer wants to do makes no sense to me. If they already have and use FAS for CIFS, and they have room to grow in both capacity and IOPS, why wouldn't they keep their data centralized (and make full use of the licenses/hardware already invested)?
Between off-site backup (if necessary), anti-virus and patching duties required to keep a Windows server protected, there is no justification in my mind for what they want to do. I've spent the last year+ ridding our environment of Windows file servers fronting legacy FC SAN. I couldn't imagine going back to that scenario.
What are they going to do when this cheap iSCSI array fills up (and everything fills up eventually)? How will this "solution" scale?