change rate

Not entirely sure what you mean by “change rates”?

If you are referring to disk change monitoring that is capable of detecting file creations, modifications, attribute changes and deletions in real-time, then this is something you need to do on the OS level.  The array is block level storage and not able to detect these types of changes.

If you are looking for statics on I/O, then SANHQ has this ability in the Group I/O and the volume I/O sections.  Here you can see I/O data (the reads, and writes) going to the group/volumes.

Joe

I would suspect that since he's asking about the EqualLogic array, that he'd be interested in the block level changes.  To the OP, this is a little more difficult than what it might first seem, as like you said, it can be over various periods of time, and it could be the same block data changing multiple times, versus a very linear different blocks changing over a given period of time.

What I'd recommend is setup some snapshots at some fixed interval you feel comfortable with.  (e.g. once a day, etc.)  Then observe the amount that the snapshot reserve fills up after each increment.  You'll be able to see some patterns develop pretty quickly.  This can be a quick and dirty way of estimating what your replicas might look like if you are interested in replication.  Although, it won't be exact because of some technical differences.  

thanks for your responses.

I probably could have worded the question better! We are looking at replication as was wondering how we can get an indication of how much data is changing (and will need to be replicated).

Is the snapshot method the best way to check this?

H

@huwy,  you should have a look at this too:

Dell EqualLogic Auto-Replication Best Practices and Sizing Guide: en.community.dell.com/.../2641.aspx

thanks. we have quite a few luns so checking the snapshots is going to be a pain!

that feature would be really useful!

Thanks for your help.

I think what you'll find is that the patterns show up pretty quickly.  Make sure that you don't have any legacy processes that might sabatoge or skew some of the numbers.  Manual file copies, defrags, etc.  

As far as the idea of having the ability to run an analysis, it seems to me it would be quite easy for them to implement.  I mean, the information on changed blocks is just metadata.  From the cheap seats, it seems like they have all they need to build in this very valuable function.