2011-01-17 11:47 AM
Are there any tools that assist/automate management of multiple volumes as a group? For example, a customer has a vfiler with multiple volumes assigned. They are paying for a certain amount of space in total. Can a 'quota' be assigned to the vfiler? If not then all the volume sizes must be known in advance or managed over their life. The volumes can be set to autogrow but this must be monitored to keep the customer under their total space allocation. It wouldn't seem logical to assign an entire aggregate to each customer.
I can think of ways to automate this with a script or two but was wondering what standard practice is (if any). Using qtrees would solve this but that is either not usable or not the considered best practice anymore.
2011-01-18 08:18 AM
if customer is paying for total capacity assigned (to total volumes), then why do you bother that? Can customer resize his volumes by himself (or is it done by you)?
if he can resize volumes, then indeed you need something, or maybe you should give him LUNs (and then you have fixed (total) amount of GB assigned to volumes, and inside volumes you can create LUNs) ?
2011-01-18 09:40 AM
Luns would help if that is what the customer needs. Qtrees also work for cifs, nfs and snapvault. But aren't there performance reasons to use separate volumes rather than split one up? All the best practice guides I've read spec separate vols for dbs and logs, etc. That is another question I'll post elsewhere. I have customers that have space for OSSV backups but the backup times cover a wide range and they sometimes want to do hourly backups. Sometimes one backup runs over an hour and messes up the schedule because while OSSV writes to qtrees, the schedule and snapshot are tied to the volume. Ideally, I'd split the backups over volumes based on backup requirements.
Customer requirements vary so we would like to accommodate their needs without excess burden on us or them. We will automatically monitor volume usage but must group all the customer's disk space to keep track. If we can set the vols to autogrow then alert us when they approach their limit that would save time and help us provide better service. The customer would receive the same information and could act on their own to keep things under control. They may not have direct visibility into all their storage assets depending on how their systems work. For example, how can an OSSV customer know when they have used all their secondary space?
I wrote a quick script to get vol usage grouped by vfiler. I'll post it as soon as I clean it up a bit. Unless I find a better tool, I'll expand it to do notifications. If needed we could turn on autogrow on vols and then turn it off on all the vols in a group when the customer reaches their limit (probably +X%). This seems like a reasonable feature set. Is this an unmet need?
2011-01-21 03:18 PM
Using Ops Manager I can see all the storage allocated to each vfiler individually and in total. This takes care of collecting all the numbers. Still finding my way around the interface. Would love to find some threshholding system for alerting when a group of volumes reaches a certain size.