1 person found this helpful
We are using SRDF in Sync and A/Sync mode, we have successfully used BCV snaps and restoring the image back to the local and remote site. BCV is also used to backup the image to tape. We tested both RDM's and vmdk's on vmfs and both worked great. Re-aligning the vmfs filesystem really help to reduce the numbers of copies that SRDF performed when syncing.
So for me its a thumbs up.
If you require further insights on our testings and results, dont hesitate to shout on the email.
Hope this Helped.
Yes. I had the whitepaper, but you know how those go... not A LOT of help for practical application.
Re-aligning the VMFS filesystem? That's something I haven't thought of... but I do it all the time for NTFS, so I suppose the same is true for VMFS. How would one do that for VMFS?
Do you mount the BCV against another ESX server, or how are you backing up an image of the VMFS partition?
Thanks for the help!
Doug I suppose it depends on your situation in terms of BCV usage. We only use to restore over the orginal, or to backup to tape. I have tested presenting to an ESX Server and that worked ok. The thing to remember on the vmfs side is that the BCV will backup the whole LUN, there you have a whole load of data to content with. RDM's give increased flexibility but may change your design parameters. We also run agents at the RDM level and they talk directly to the SAN to quiesce for particular applications.
Also BCV's cant be presented back to an ESX server that connects to the LUN you just backed up. Much like other system you cant have too volumes being presented with the same signature header to the same ESX (or group of) servers.
We are using SRDF in Sync mode, replicating our LUNs between our two main buildings. We have distributed our ESX Servers also between this two building and grouped in pairs into our farms. We use DiskMaskLUNs to mask the R2 to the DR ESX server.
We have simulated a disaster, and we have done manually a one LUN failover from a Windows server using EMC Solution Enabler, unmasked the LUN, rescanning SAN, registering the VMs and try to start the VM. It worked.
Actually we are backing up our VMs like physical servers, using a backup agent.
Thank you all for the help.
I have some more reading and design work to do
Doug I suppose it depends on your situation in terms
of BCV usage. We only use to restore over the
orginal, or to backup to tape.
I think I see the source of my confusion: we don't have a way to write the BCVs to tape without an intermediate host. For our Exchange server, we snap the BCVs, then mount them on a 'backup' server that runs a Networker client to backup the files. With VMFS, I can't really present the LUNs to a Windows box and expect to get anything useful to backup
I've seen the posts regarding VMFS partition alignment and your note that "By default, ESX server will create VMFS that are misaligned". My question, then, is how do you correctly align the partition? I'd use something like diskpar on Windows to accomplish this goal...
I'm also looking into BCV solution for backup. We have an ESX box dedicated for backup where the BCV's are mounted and backup the vmdk file from there. My question is do you need to "deport/import" the LUN's before and after the BCV process? If so, how do you "deport/import" the LUN's?