The plan is to retire our existing Clustered Windows 2003 file servers and HP EVA SANs by taking advantage of the CIFS functionality of the FAS3270s that we've purchased. The existing structure has different groups, Training and Engineering sharing a single drive on the file server. The headache is obvious, that as the different departments consume their storage they also impact the other department and it becomes a problem tracking which department is actually the one at "fault" Being new to CIFS on the Netapp my initial thought was to create a single volume for each department and migrate their data to it. The push back has been that this will result in ore administrative overhead maintaining multiple volumes and staying on top of which departments need to be grown. The other admins would like a couple big volumes with all the data stored there.
Figured I'd ask the group, see what the rest of you are doing, and any gotchas to look out for.
Having not used the netapp's quota functionality, I'm assuming I'd setup quota's by each group, so that Ops could have 500 GB, Engineering 1 TB, Maintenance 200 GB, etc and if a single group hit the limit they would be stopped from consuming any more space, and not impact the other groups. The quota would use the Active Directory groups, or would I have to set up groups within the NetApp itself? What about shares that have a couple different AD groups using them?
Quota's have always been a sore spot with Management here, typically they buy into it until the first time a hard limit is reached and then they panic and insist the quota's be turned off rather than having the group do some cleaning.
Re: CIFS - Multiple smaller volumes or one big one?
We actually used Northern Parklife on our cluster servers, never could get it to work right on the cluster. Seemed to struggle with fail-overs, and would just stop working. I understand they've changed management since then, so may be a different company now.
I like the benefit of maximizing the rate of deduplication, hadn't really considered having more data to deduplicate would probably result in more "hits" then if each Department was on it's own volume.