Data Domain replication or stick with Veeam Backup Copy Jobs

[bpayne]

The main question is should I change and use Data Domain Replication or stick with Veeam Backup Copy jobs and suffer (I'm already using Veeam BCJ)?
First before you respond with the we do not support Data Domain storage replication, I know this, but please hear me out below.

Back story - Currently, we use Veeam Backup Copy jobs to "replicate" data from our PROD site DD4500 to our DR SIte DD4500. Over the years, I've had a huge amount of growing pains using Backup Copy jobs from DD to DD. Yes, I understand the underlying issues of DD and how we are using it as a backup repository, but the design and investment has been made and this is what we have to use for now.

Key issues with using Veeam Backup Copy jobs with Data Domain:
1. Over a thousand VM's, some very large ~8+TB
2. The full merge seems to look like "new" data that the DD has never seen. So I have constant space issues on Data Domain. For ex - for a single day our pre-comp written is 1.5 petabytes!
3. Copy Intervals expiring b/c #1, so we don't always have restore points for all vm's for that day. It may take an additional day
4. Extremely SLOW copy bandwidth. Something like an average of 1-25MB/s. Yes I know this is b/c of Data Domain re-hydration
5. Over time, usually like 6 months seems to be the sweet spot, I have to completely delete and 'start over' all Backup Copy jobs b/c of #2.
6. I am only actually able to copy like 7 restore points for each BCJ (our general retentions are 35 restore points, 13 montly's, 2 yearly's), b/c of #2
7. I have to create multiple BCJ's with multiple gateway servers assigned so I can meet my Full Merge times within that 24 hour period. So this is additional administrative overhead to configure and monitor

Given all those issues, I am considering changing to Data Domain replication and dumping the Veeam BCJ's. I have scoured the Veeam forums and it seems like most Veeam/Data Domain customers are using Data Domain replication despite no support. However, I've also seen issues posted in Veeam forums with customers using DD replication b/c of performance issues.

I'd like to here what all you Data Domain customers are doing for your backup replication. Even if DD replication solves half of those problems above, I am highly considering...

Thanks.

[chjones]

I had a very similar setup to you, but using HP StoreOnce instead of DD and also fought with similar issues.

In the end, I used the StoreOnce's inbuilt replication engine to replicate the CIFS backup share to another site. The inbuilt engines with their much smaller dedupe block sizes made replication much more efficient and bandwidth utilisation was also above what Veeam could push.

The downside is Veeam is not aware of the replicated data. But Veeam is very quick to rescan a repository and catalog the restore points so if I needed to restore from the replica I'd just add the remote CIFS share to Veeam, it would pickup all the restore points, then restore with ease.

I know Veeam will say they don't trust storage replication as Backup Copy Jobs re-write the restore points so you are sure the backup files are not corrupt (storage replication will just copy bad data). But it's really up to what works best for you and what risks you are prepared to live with.