Where to store replication metadata with Veeam 6

[chrisknight]

I have searched a bit but I just don't have a good grasp. We have been running Veeam 6 for a while now but it was upgraded so all of the existing jobs were upgraded too. The replication job (after the upgrade to 6 from 5) always stared "Legacy Replication" job. It always worked before so I never bothered changing it. I upgraded to Veeam 6 patch 3. I deleted all of my jobs to recreate them. The backup jobs seemed straight forward. We have 2 NAS boxes where we backup all VM's every other night to every other NAS box. Backing up to NAS on LAN so reverse incremental, LAN target, VSS, etc... When re-creating the replication job I was a bit confused.
We have an HA cluster with 2 hosts. 2 iSCSI boxes attached to both hosts. One for main VM's to run, one in a vault where we send the replications every Sunday at same physical site. I was surprised to see that the default replication storage location when configuring the replication job was the 2nd iSCSI by default which is where I want it to go... Built in logic to choose the best location? Anyway, I was asked a question where I want to store my metadata for the replication job. I had only 3 choices. The 2 backup NAS repositories, and the Veeam C: drive repository. I chose C: Veeam server repository because I didn't know what else to do.
There are 10 VM's split (50/50) between 2 ESXi 4.1 hosts in a HA cluster.
What should I do here?
How much metadata is there? (c: drive consumption)
Did I do the right thing?

Thank you,
Chris

[Gostev]

You can use any backup repository that is physically located in the source site. A few GB of free disk space should be plenty in your case. Thanks!

[cmdrriker]

But what happens in chrisknight's scenario if the veeam server fails, taking the metadata repository with it? Wouldn't that destroy the ability to do restores or replica failovers to multiple restore points?

Where is the best place to put the metadata to insure it's availability? Why isn't it automatically stored on the target datastore along with the replica and restore files?

[dellock6]

I think because replica is started by the source proxy, not the target one, so there is probably some activities that this proxy has to do before sending new data to the target proxy.
About the metadata, it requires a veeam repository, so you should apply the same security measures like any other repository; what if you loose a repository holding all your local backups? If you are already using a repository for your own backups, go and use it also for replica metadata.
You are not forced to use the default local repository created during Veeam installation.

[Gostev]

If the metadata is lost, it will be automatically recreated by the job, resulting in a longer incremental run.

[jsprinkleisg]

I understand there is some performance advantage to putting replication metadata at the source site. Can Veeam still failover replicas if the source site is down and replication metadata is inaccessible?

[tsightler]

The metadata has nothing to do with failover. It's basically just a place to store the hashes for deduplication during replication. Since dedupe is performed by the source proxy it's best to keep it on the source side, otherwise the metadata has to be read/written across the WAN by the source proxy.

If the metadata is missing, Veeam will simply recreate it by rescanning the target side and recalculating the hashes, but it's only needed during a replication, not for failover.

[mdornfeld]

Thanks Tom. This was exactly the crystal clear answer I was also looking for on metadata and failing over.