Replication Site to Site

[Rohail2004]

We’re in the process of building another datacenter, and looking for a Replication software to make sure some of our most critical servers are getting Replicated to another datacenter on daily basis and have the up-to-date data in case of our primary datacenter failure. We have been using Veeam backup and loved it, but haven’t used Veeam Replication for site-to-site. How is your experience with Veeam Replication between site to site? Did it work properly in the event of site failure? Any suggestions/advice would be highly appreciated.

Thanks.

[Vitaliy S.]

We've got plenty of existing discussions and recommendations, please look through them: http://www.veeam.com/forums/search.php? ... N+ESXi+ESX

[zanderins]

I hope this is of value.....you absolutely need a great deal of bandwidth for the replication to succeed. Why do I say this? The one true downfall I'm finding with Veeam is not the applicaiton but the size of the files with backups and replications. I have one server that is 450GB total size and all together I have over 2TB of server data that gets backed up nightly ( I also do a replication of the servers as well). I first tried the replication by putting the "replication" server at an offsite location and I found that my 3MB (and I tried our 5MB circuit as well....we have redundant circuits) could not handle the replication processing.....I was "pulling" the replicated files from the hosted data center and "nada"....wouldn't do it....failed. I tested many options and ultimately created a 3rd backup job that is an incremental backup job only and transferred the initial backup to a USB HD (for all the servers) and then "seeded" the data at the offsite data location and run a "sync" of the VIB files nightly (which is about 40GB a night of changes for each day Monday thru Friday). The key is to set the "Full Backup" to only occur 1-time a year at which I have December as my 1-time full backup at which I will then transfer all the Full Backups to a USB HD for archiving and start all over again with "seeding" the backup data for off-site Disaster Recovery. Again, replication won't work if (1) you have huge amount of data; and (2) if you have limited bandwidth.....you truly need a 10MB circuit but ideally you need a 40MB circuit to really rip the data across and be productive. From a cost standpoint, the incremental backup method and then seed the job offsite and do a nightly "sync" of the data (grab changes only at which the most recent VIB file will be copied) is the most economical fit from a cost standpoint (won't break the bank). I hope this saves you some time because the testing took forever for me due to the amount of data we have.

[Rohail2004]

Thanks for detailed information. We're going to have a gig line between two Datacenter..

[larry]

I replicate between sites using 100 meg. I also backup between sites never sending tapes on the road. Some VMs have 1tb databases, still no issue. The one thing I would recoomand is have a Veeam server at the remote end pull the backups and replica's this way during a DR you have all the restore points you may need. I have two VMware setups with two VC's, one at each location, this is for DR, Veeam can replicate, copy and backup between them. I am also in installing a second veeam server at each location to help with timing and extra DR, you license the esx servers not how many veeam servers you load.

[st0623]

In monitoring this, folks, I want to weigh in on using the HyperIP virtual WAN optimizer to reduce the bandwidth. We have significantly reduced the required bandwidth for Veeam replication. In most cases, we can reduce the requirement by 50% over a WAN link. Let me qualify this, a metro GigE is going to be sufficient between data centers, but what if you can't afford a metro-GigE? What about using HyperIP to optimize Veeam replication and see what pipe you truly need. Maybe with HyperIP that 10Mbs pipe would look like a 20Mbs pipe or even a 40Mbs pipe to the app. We're doing wire speed or nearly 95% of that, by getting rid of the effects of latency and packet loss, then doing block compression on the payload. That means you can leverage a cheaper MPSL or Public Internet link for replication at half the cost of a fiber link AND leverage HyperIP's cost savings. Eval it at: http://www.hyperip.com