Replication Failover and Failback

[mik3yl3]

Case # 01902652 -

Hey Guys,

We have been testing Veeam replication and its failover/failback feature. Can i ask what have your guy's results been? your thoughts and maybe best practices? We have had horrible results. We have a large datacenter in Boston and in San Diego with a 1 GB link. We tested using a stock Windows Server 2012R2 box, replicated from Boston to San Diego (50-60 GB), failed over and failed back with WAN accelerators. The failover took about 30 minutes, failback took almost 2 hours. Then we did it going the opposite direction and length of time was the same. The setup is fairly easy and i dont think im missing anything during initial configuration. I notice a ridiculous amount of snapshots done on each side to do whatever verification process it's trying and probably adds to the time. Any thoughts would be appreciated.

[foggy]

Longer failback times are expected, please see the explanation at the beginning of this thread. The number of snapshots taken is explained in the corresponding user guide section.

[DrWhy]

I think you'll find this thread helpful as well.

vmware-vsphere-f24/feature-request-spee ... 28233.html

[kenny782]

Its only 25 GB. This is a test server. The initial seed to my DR site took 3 hours.

I failed over, put a test file on the desktop and am now failing back...its going to take another 10 hours...i'll update when done.

Something is not right...

This post is basically exactly what I've been testing including size and times.
I do have a proxy on both ends so based on some of the replies I'll play with the proxy settings on the job.
I've been testing everything in house before shipping anything out on different VLAN's so the connection couldn't get any faster!

Support has been ZERO help, after several cases and seemingly wasting my network admins time working with them I get the same answer "these failback times are normal".
I knew if that were true VEEAM wouldn't have any customers..... Apparently the community forum is far more helpful that the support engineers.

Anyway thanks for the suggestions guys.



What I can't understand honestly is I understand you want to verify integrity so you don't corrupt VMDK's.
But if Production is up when you failover and allow VEEAM to shutdown the server
It creates a snapshot....replicates that snapshot over. Assuming the original snapshot is intact in production I would think it should be able to just analyze the subsequent VMDK's generated by the snapshot calculate that and then replicate that data back.

Maybe I'm over simplifying but it seems logical to me....
Now if I had to failover with production down THEN I can understand a lot of analysis required to failback.

Just my two cents but I'm certainly not a software engineer and don't understand the internal workings of VEEAM to the point where I could say anything I just said is correct.
It just seemed logical.


I'll post back what the results are after I play with the proxy settings.

-Kenny

[DrWhy]

The primary reason this is slow is detailed here. The good thing is that it's on the dev's radar. The bad thing is that it's been on their radar for while now.:

vmware-vsphere-f24/feature-request-spee ... 28233.html

[foggy]

The good thing is that it's on the dev's radar. The bad thing is that it's been on their radar for while now.

And if you look at the last post in that thread, we've finally hit that target.

[DrWhy]

boom! thanks!