Replication Issues

[prutter]

Hello All,

Pardon me if this is the wron forum for this topic but her it goes.  We recently setup a DR site so that we can replicate all of our data.  We have an IBM N3600 (aka Netapp 2050) on the prod side and a Netapp 2020 on the DR side.  We run ESX 4.1 on 3 NFS Datastores.  We are running Sapvault for all of our CIFS data and Snapmirror for all the VMware data. We are replicating only about 25 servers and VM's and have a concern about the ammount of data that is changed daily. On the VMWare side we have roughly 1.5TB of used space and we replicate about 55GB of data per day.  That is 55GB worth of changed data on servers that don't have a lot of data changing ( i think).  On the CIFS side we have about 1TB of space and we snapvault about 30GB per day (changed data).  We are finding it very hard to believe that we have that much "CHANGED DATA" per day and as you might imagine, the amount of time it takes to transfer all this data to the DR site.

My questions to everyone are this...Is there a really good way to find out what servers/data that is changing daily so we can troubleshoot this issus?  Is there an easy way on the CIFS side to do the same?  We may need to modify the way we have things in the office but don't really know an easy way to identify the data.  We are a small shop of 110 users and don'thigk this much data should be changing.  The pipe we have between our DR site will kill us in cost if we have to upgrade the pipe size for more bandwidth.  Any thoughts or comments are welcome.  Thanks!!

Perry

[yannbizeul]

This is a very NetApp Specific issue I think. Did you post on Netapp communities ?

On way I think of to get a more VMware centric view of changed data is to take a snapshot of all VMs, let it run for the day, then check the amount of space used.

Do this on each NFS datastore and you should find which server, if any, is changing so much in one day.

The storage view, in vSphere, is pretty convenient to get a view of snapshot space.

[whynotq]

as it's NFS and replicating at the File level it may be the transfer size or file data changes that are skewing the volume. As previously suggested I'd get talking to NetApp to see their take on what's going on.

Out of interest do you back the VMs up as well as replicate the file-shares? if so the incremental backups can give a good guide to daily change rate and in some instances can cause the increase if the replicating software sees changes to the file archive bit (backup exec does this) and takes this to be file change so replicates.

regards

Paul

[prutter]

yannbizeul

That sound like a decent idea.  I've been moving vm's into a datastore that doesn't get replicated to see that way but it's a long process.  Thanks for the suggestion.

Another question i have about the snapshots.  I took snapshots of all servers in the datastore to see which one is causong all the data but more confusions sets in.  I  see some VM's snapshot delta file has changed by over 1GB but when i look at all NEW and MODIFIED files for the past few days i don't dome anywhere near that file size.  In fact the size of all the changed files is less than 100MB.  Anyone know why the delta file is so much bigger?  We are trying to figure out why we have all this changed data getting replicated.  Any help is appreciated.  Thanks again!

[DSTAVERT]

You might want to look at some of the backup/replication tools that take advantage of VMware Change Block Tracking. There are usually Trial copies available. Are you using any of NetApps dedupe technology?

[prutter]

I will look and see what tools might be available but if you have a couple ideas i wil listen.  It's really hard to pinpoint exactly where all hte data is coming from but we have a 3MB pipe to the DR site and we still can't repicate all the data.  I have one datastore that we snapmirror that changes by 30GB each day.  I've moved all the swap file to a differernt datastore and that seems to help but stillhave a long way to go.  Any other uggestions are welcome. Thanks for all the replies!