New Veeam Setup Options

[paallergy]

CURRENT SETUP

2 ESXi hosts
1 SAN (direct connect - FC)

6 TB of data
main application uses SQL Server
80 users

Backups

• not Veeam
• backups done at night

NEW SETUP

new hosts (ESXi 6.5)
new secondary site that can be used for DR (100 Mbps)
30-minute RPO
(If possible. Concern is that users will notice.)
daily change rate: 700 GB (per Veeam One)
Veeam for backup and replication

• some VMs will be replicated
• all VMs will be backed up (for longer term retention)
• backups will be verified (SureBackup) at secondary site
• backups will be copied to removable media at secondary site (offline)

OPTION A: 1 host, local storage, all flash
OPTION B: 2 hosts + witness (vSAN all flash direct connect)
OPTION C: 2 hosts, FlexPod Express (NetApp hybrid)
OPTION D: 2 hosts, FlexPod Express (NetApp all flash)
OPTION E: 2 hosts + SAN (direct connect, no storage integration with Veeam)



QUESTIONS

1. Which options above (A-E), if any, will allow us to do replication and backups during business hours without users noticing (no stun, freeze, etc...)?
   • For options C and D I'm assuming application aware NetApp snapshots
     then either Veeam replication to secondary site or NetApp snapshot mirroring (see #4)
2. Which options above (A-E), if any, should absolutely be avoided if users shouldn't notice replication snapshots?
3. How does VSAN (all flash) compare to NetApp storage integration for snapshots (backup, replication)?
4. Which is more efficient for replicating VMs from main site to secondary site?
   • Use Veeam for replication
   • Use NetApp snapshot mirroring
5. For NetApp storage snapshots, is there a noticeable difference between hybrid and all flash?
6. Anything I'm missing or should consider?

Thank you!

[PTide]

Hi,

Freezes/stuns may occur due to a large snapshot commit. It cannot be guaranteed that users won't notice, especially for highly transactional VMs. However, the issue can be largely mitigated via the usage of a faster storage and storage snapshots integration. That applies both to backups and replicas. So I think C or D should be the way to go.

Which is more efficient for replicating VMs from main site to secondary site?

That depends. Please check this thread for considerations.

Thanks