HPE Storage Users Group

Hi, What local copy feature can one use to copy data from one array to another. Would it be one of the peer* licenses?

Basically we have 2 arrays in a remote copy config in separate buildings and the target array is to br decommed. We were wondering if we could move the r copy data from the to-be-decommed array at the block level to the new array then re-group the new array with the source? The other option is re-seeding from source to new array but this would take over 110 hours.

The other idea I have is to daisy chain RC so that we have source > repl1 > repl2. repl1 and repl2 are in the same comms room so the seeding would be quick but not sure if this is supported or even possible as async schedules could overlap.

Thanks

Ideally you would co locate the arrays to provide the initial seeding and then physically move the new array to the remote site. Peer motion at the DR site isn't really intended for this and would likely complicate the process. There are any number of ways to transfer the data but it depends on your SLA's etc.

If you can afford to suspend the replication for a short time per DR group then I'd take a look at the remote copy user guide, page 123 for 3.1.2 MU2 "Using Tape for Initial Sync". Appreciate you don't want to use tape, but the seeding process through snapshots will be the similar, just minus the backup and restore as the data can be remote copied from the DR to new array and a variant of the above process used.

The biggest problem is if you get the process wrong then you're likely going to have to do a full resync anyway, so some element of testing and a rollback plan will be essential.

Hi Cleanur,

Yup tape restore did cross my find but not sure if our backup software supports block based backups. In the mean time I have raised a call with HP and the guy says that doing array1>array2>array3 is doable.

The question which he could not answer was what happens when array3 is fully synced and you are trying to pair up array1 > array3 volumes....will RC 'see' this data as fresh data and seed from scratch? That would be bad.

In a way its a bit like restoring from block-based backup onto array3 so you think it might work?!

Cheers

The tape method is ugly because the backup needs to be block based if I recall and very specific process so the data lands correctly on the other side. We just let things replicate, took a few weeks, but was far easier than trying the tape method.

Hi,

Im not suggesting doing the tape method im simply saying that seeding from [array1]>[array2>array3] then taking array2 out and syncing array1 and ray3 with RC may just be the same as doing a tape restore with the -nosync switch.

thanks