Ideal Datastore Configuration for Replication

[victorcharlesedwards]

We're new to Veeam and would like to setup B&RE on two active ESXi servers (i.e. A replicates to spare datastore on B, B replicates to spare datastore on A), but are unsure if this is a good approach, and if it is, what's the best datastore configuration (and RAID) we should deploy for this setup. A standalone shared storage target (SAN, NAS) is unavailable in this environment.

Each server has seven 3.5" 3TB (7200RPM 6Gbps) drives, we are considering the following RAID configuration to maximize performance, increase redundancy, but also to isolate Veeam replication data from the primary datastore.

Server A

• - Four drives as a RAID10 (Server A primary datastore)
  - Three drives as RAID5 (veeam replication datastore target for Server B)

Server B

• - Four drives as a RAID10 (Server B primary datastore )
  - Three drives as RAID5 (veeam eplication datastore target for Server A)

Is it overkill to have separate RAID volumes / datastores for replication data? Is there a better approach given the parameters of this project (no shared storage target)? Thanks!

[dellock6]

Hi Jon,
Veeam replicates VMs at the vmdk level, not at the datastore level, so there is no need to have the same configuration in both ESXi servers or to have the same size of both datastore to replicate between them. Size yuor datastores based on the VMs more than the datastore themselves.

Luca.

[victorcharlesedwards]

Thanks Luca,

Each server will host 2-3TBs of VDMKs over the next 18 months, but we want to provide the most space available with the least impact on performance, so I think we'll just do a 6 x 3TB RAID10+1 on both servers.

[dellock6]

Raid10 is a good choice for sata disks, especially on low-end raid controllers with little or no cache. You have no write penalty and performances are the best you can get from that kind of disks, at the cost of loosing some disk space. A single large volume will give you more speed (more spindle disks running for the same raid group) and also more flexibility (large datastore to help you with vm growth).

Luca.