What are my options to restart replication ?

Well, sounds like to me that it hasnt replicated for a year you have no choice but to re-replicate everything.

Is the horrible performance on the source due to using network resources to copy?

If so, you might want to set a bandwith limits during business hours.

Another option would be to peform some sort of "tape silvering"  (I believe is the correct term) and backup your data to some medium, send it to the DR site and then restore.  Then restart your replications.

Being overly conservative however, I would choose to simply restart replications all over.  On our new arrays, we have started dedicating a 1GB port simply for replication so we do not compete for network bandwidth.

hmmm,....

interesting...

Have you set your replications on the source to PAUSE/ STOP.. dont recall what the option is and then bring your DR site back up?

You might then be able to control the replications by starting them one at a time until they get caught up.  (That was recommended to me by EMC when we perform our DR testing).  Dont restart all the replications at once.

Again, it sounds like once that DR connectivity is online, its trying to resume.  I'm surprised that the SavVol hasnt been consumed by the replications being down for so long (assuming the replication sessions were still "on" but the connectivity "down").

DHoffman wrote: Is the horrible performance on the source due to using network resources to copy?

nope, have not even gotten that far ..just fixed LACP trunks on DR celerra and as soon as i did that source Celerra performance went to crap (pardon my french).

unfortunately this is a very performance sensitive customer (modality sending X-rays/CT scans) so can't afford to try/guess, got a SR open and waiting for an action plan.

Hi,

I don't know if it's too late, but...

I would stop and delete these replication sessions, then restart them one at a time, and monitor the DM resource usage, specially CPU utilization.

I had similar cases, and on very busy systems needed to restart them on weekends, or low I/O periods.

Also, the replicatorV2 usually don't create so much load, so I don't understand why you had the performance hit, unless the system is already overloaded.

Please, share your findings.

PS: as said, the newer codes have lots of improvements on this area.

Gustavo Barreto.

the case is still with support.

Gustavo,  i assume that i will need to re-enable interfaces on target Celerra. That's the issue, the minute i re-enable interfaces on target Celerra ..performance on source Celerra goes to toilet, customers start calling that it takes 10 minutes to pull up an 10MB x-ray file.  That's without replication even running.

Hi,

it's weird...

You can stop then delete the replication on the souce side with "-mode source" option, then do the same thing on the destination side with "-mode destination" option.

This should stop any traffic, and delete the internal replication checkpoints on both sides.

Then, recreate the replication sessions on off peak times...

Gustavo Barreto.

ahh, so if i use -mode ..it will not try to talk to remote Celera, it will simply delete replication relationship ?

cool, so once i run through these on source and target ..i can re-enable interfaces on target Celerra and re-create my replication sessions.  Probably a couple of file system at a time ..so not to bring production box and replication link to its knees. Thank you so much Gustavo

Hello All,

Can i get any documention to configure the replication between Celera to VNX.  I have configured the same however i need to submit the document to the customer ... I have downloaded one doc but it was content protected so cannot copy any data from that... If anyone have the document kindly inbox me at ragavs2908@outlook.com

Thanks in advance,

Ragav Sidhu

Why not send the customer the official VNX Replicator documentation ?

Yep – documentation or white paper

If you want to have a custom documentation for your customer then you would have to create it yourself