Nimble Backup fresh setup - Best practice without Enterprise Plus license

[stekkew]

Hi all,

Despite reading a lot of forum posts, veeamBPs, HPE whitepapers etc., I still have some questions regarding our own setup and how to implement Veeam in a good way.

We recently upgraded our production environment, which consists now of a stretched VMWare cluster (4x Proliant DL380 Gen10+) and 2 Nimbles (HF40 - iSCSI setup) with synchronous replication enabled.
We have about +-80 active VMs, 30TB of data (about 30-40% are images/video archives which don't change often).

Our existing Veeam setup consists of an all-in-one Veeam VM as Veeam Server/backup proxy, which uses Virtual Appliance mode (Hot Add) and writing to a Synology RS3617xs+ as main (NFS) backup repository.
We also have another Linux server with iSCSI storage (older Synology) for immutable backups and a HP tape library connected to an old server for Backup to Tape.

Also in the project of renewing the hardware and to simplify the management, we are setting up our Veeam environment from scratch.
We recently acquired an HPE Apollo 4200 to:
- Use as primary backup repository (ReFS volume)
- Use the Apollo as backup proxy
- Use as Tape Server (FC connectivity to Tape Library)
We will be building a VM as management server as well, and as a best practice both machines won't be member of our AD

At this moment we only have a Veeam Enterprise License for those 8 sockets (Not Enterprise PLUS, so we cannot benefit from storage snapshots).
On the apollo we have 2x 2-port SFP+ adapters, which we CAN connect to our normal network + iSCSI storage (Nimble) network, but here things become a bit fuzzy.

It's a bit unclear to me if we can benefit from connecting the Apollo to our iSCSI network, without having any issues (without storage snapshot integration), corrupt Nimble Volumes or VMWare Datastores, and still use VMWare snapshotting?
I think so, but is it a good practice to do so?
I assume we then have to add our Nimble in Veeam (Storage Integration) - or is this specific for use with storage snapshots? Or should we install Nimble Connection Manager (NCM) on our Apollo Server to be able to connect properly to our Nimble through iSCSI.

Or will we always be needing a Veeam Virtual Appliance in Hot Add mode for transfering data to our Apollo Server (which would be through our "normal" network instead of iSCSI)
At that point the CPU and RAM of the Apollo is not under stress at all, it is the VM that is doing all the work... Whereas I thought, it would be the Apollo that would be the "workhorse" (backup proxy) for those primary backup jobs.

Thanks for your feedback and pointing me in the right direction.
I can provide more information if you need to.
Kind regards

[HTownVinny]

Stijn,
I have experience working with many of the specific products you mentioned so I want to ask a few questions before suggesting anything.

1. What hurdles do you have to overcome to attempt a configuration that includes adding additional storage to your deployment? In other words, if you own the hardware mentioned and you have sufficient licensing, why not connect everything and run jobs targeting the additional storage?
2. In my experience, NCM has only helped situations, I have not had any negative experiences having it installed. Is there something that would prevent you from installing it where appropriate?
3. I read that your using both iSCSI and NFS. Is there a reason for using both instead of focusing on 1 protocol? I ask this question because having both in place creates a variable that could make troubleshooting more difficult. Simple & Secure seems like a better setup to me.
4. Which version of VMware (6.x, 7.x) are you using? Unless I missed it, I don’t see that detail. That detail may play a role in how best to adjust your configuration. Each version has its own capabilities/limits.

Speaking from experience working at a VCSP, it seems that you have significantly more available to you than most. That being said, I would suggest you focus on simplifying as much as possible and then troubleshoot performance from there. As long as you keep security in mind & don’t impact the business negatively, More backups are better than less.

-HTownVinny

[Andreas Neufert]

Implement 3-2-1 backup rule or the enhanced one.

Implement the Apollo 4200 as hardened repository to have immutability. Not sure how you do immutability on your old synology. It would only work if you cut the management completely out of your network for this box and do a direct connection between the immutability box and the backup server + leverage hardened repository as well?

Anyway use HotAdd/DirectSAN to backup to Apollo 4200 hardened Repository, then create backup copy jobs to another site (can not burn down at the same time or get fludded at the same time) to maybe your synologys or another Apollo.
Implement fast network in between (min 10GBe). Place the B&R Server on the second site so that you could failover to that site. Place some VMware resources there so that you could do instant restore for your customers for the most crictial environments.
Maybe place the tape server and the tape library at a thrid datacenter and take some of the backups out of the library at a weekly base at least (offline air-gapped media).