Nimble and Veeam - Best Practice?

[oliverL]

Hey there,

we have different Priorities for virtual machines in vmWare where each priority is given different retention times for the backup (eg. Prio 1 = 1 Snapshot on the local nimble, 4 weeks of snapshots on the replication nimble and 8 weeks on the Repository Server).

I would use Tagging in vmWare to define which VM has Prio-1, Prio-2 and so on, so that the VM with the Tag Prio-X gets into the corresponding Veeam Job.

Then i would use these Jobs in Veeam to make a Schedule for Prio-1 where the First Backup goes to the Primary Nimble-Snapshot and then as a secondary Destination the Replication-Nimble is assigned.
Then i would need to create another Job in Veeam that uses the replicated Nimble-Snapshot as a source to get the Snapshots from the replication-nimble to the Repository.

Makes sense right?

Since VMs with different Priorities can reside on different Datastores it wouldn't make sense to create many Volume Collections on the nimble right?
So would it be better to put all LUNs/Datastores/Volumes into 1 Big Volume Collection since Veeam will be steering the Nimble Replication and Snapshots?

Hope somebody can enlighten me a little bit, because i couldn't find a "best practice" for Veeam and Nimble...

[PTide]

Hi,

Seems good to me, apparently you've already read this guide Nimble has its own best practice pdf, worth checking.
Regarding Volume Collections - I'd try to place all VMs that have the same priority onto the same datastore and create a separate Volume Collection for each Prio group.

Thanks

[oliverL]

Hey PTide,

thx for this. Yeah i've read the first Guide but it won't work...

I've opened a ticket, but the whole text isn't formatted anymore
#01997612

:Edit: Unfortunately the Nimble-PDF isn't telling anything about the Snapshot functionality as it only describes how to attach a nimble volume to a Veeam Repository Server as a backup target

There is something which is unclear for me in the Veeam Blog-Post:

One of the key benefits you get from backups from the secondary storage is minimizing of the performance degradation that impacts your production storage during the backup process.
To be able to back up a VM from the secondary Nimble array, you need to create a snapshot on your primary storage and replicate it to the secondary storage first. After that, Veeam Backup & Replication will back up the VM from the secondary storage.

Does the nimble have to create this Snapshot and replicate it or can i use Veeam to create this Snapshot?
- If i can use Veeam for that. Do i need to create a Replication Job first and then Append the Backup-Job to it? Or can this actually be done with One Backup-Job?
Somehow i miss this Information in every blog-post or help-file.

[PTide]

You can use Veeam to create Nimble snapshot, please check this article (it describes how it works) and this one - it describes the whole configuration process.

P.S. Actually QA team advised that it is possible to do everything in one Veeam job. You need to check "enable backup from storage snapshot" checkbox and "enable failover to primary storage snapshot" in job settings, then you should specify both primay and secondary Nimbles as a secondary destination and click on the secondary Nimble entry and check the "use as the data source" checkbox. You can also adjust snapshot retention settings there.

[oliverL]

Can it be that the Blog Post is totally outdated?
https://www.veeam.com/blog/nimble-stora ... guide.html

My Backup-Repository is a Scale-Out Repository from the Backup Proxy which also sees the Nimble Volumes via iSCSI
I also configured secondary Targets where i added Nimble Snapshot and Nimble Snapshot Replicated Copy. The Replicated Copy should be used as a data source.

So far so good.

But Veeam doesn't like this.
When i start the Backupjob the third thing which Veeam tells me is "cannot find matching replicated copies of Nimble Snapshots" and if i wouldn't tick "Failover to primary storage Snapshot" Veeam wouldn't do anything.


Sooo... is the blog Post outdated with the GA Release of Veeam 9.5?
Somehow Veeam wants to look for the replicated storage snapshot first instead of creating the primary storage snapshot, then replicate it and then use the replicated copy as a data source to do place the backup files on the Scale-Out Repository

In the VM i get the Warning "Failed to prepare VM for processing from secondary storage snasphot, failing over to primary storage snapshot"



Edit: If needed i can open a ticket, so that i do a remote session with a Veeam Engineer to get a look at these jobs.

[PTide]

Yes, please open a case with support team so they can find out the reason why Veeam has not detected a snapshot on the secondary storage. Please don't forget to post case ID. Regarding the blogpost - QA team confirmed that it's still relevant. Could you please be more specific on what exactly did not work?

Thank you

[oliverL]

Case # 02016715

About the Blogpost:
What isn't working as described in the Blogpost is the part where Veeam uses the nimble replication snapshot as a data-source for the Backup to the local repository.

I've enclosed Screenshots in a word-document into my case - maybe this will help you.

[foggy]

It should work like described in the blog post, so please look into the possible issues with the help of your support engineer. Thanks.

[oliverL]

Yeah i know - it worked already but somehow now it doesn't. Hope that your support can help me.

[oliverL]

Update:
Support need to get in contact with other support engineers to see how the nimble storage integration works I've also had to show this forum post to the support engineer because he didn't believed me that Veeam should do the replication-stuff which we where talking... (yes i've included a link to this thread into my ticket)

[foggy]

Thanks for the update, Oliver. Unfortunately, I cannot make much sense of the case notes, since it is in German, so I'd appreciate if you continue keeping us posted about details of your investigation.

[oliverL]

Ticket is going to get internally escalated since the support assumes that Veeam has a Bug.

[oliverL]

Seems like there needs to be a replication of the volume collections configured on the nimble first before you try "use backups from replicated snapshot". Or else Veeam won't find the replicated snapshot copy on the secondard nimble...

Which means on the other side i have the replicated Snapshots from the Nimble as well as the Snapshots from Veeam. Uncool but atleast it seems to be working now.

[foggy]

Yep, this is a valid note that should be clear from the documentation. Thanks for the heads up.

[oliverL]

Jep the technical Writers also did some overhaul to the Helpcenter-Page:
https://helpcenter.veeam.com/docs/backu ... tml?ver=95

Now there is standing that you have to create a Volume Collection Replication before using this Backup-method and the whole page get alot more Information!

Really nice!

[bjdboyer]

Is there a best practices paper or document on using storage integration, especially Nimble? Like organizing the VM's to the datastore/volume with the backup jobs? Our VMware guy was at VMworld last week and wants to go back from VVOLS and start using storage snapshots. But question is the best way to design the VM->Datastore->Job relationship to best utilize the snapshots.

TIA,
Bill

[DGrinev]

Hi Bill,

Please read this existing discussion as it contains a plenty of useful information and links regarding integration with Nimble. Thanks!

[bjdboyer]

In our environment we only have the 1 Nimble as primary storage for VMware. The Veeam repository storage is on a separate IBM V3700. We do not use Nimble replication for offsite as we don't have one offsite. We use SRM and soon to be Zert0 for replicaiton and failover to our d/r site. There are backup copy jobs to send the backups to a couple different offsite repositories.

So we are trying to leverage Nimble snapshots to minimize the effect of the snapshot on the VM's for the backups. The proxy servers have the datastore LUNs mapped to them for SAN transport on backups. we do not retain any volume collections on the Nimble. Space issues mostly.

So my question was how to organize the backup jobs with the VM placement on the datastores to get the most of using Nimble snapshots just for the duration of the backup. Would we create multiple datastores for each customer/backup job so that the job could do all of the VM's in a single snapshot without having to delay any other job. Right now we have a couple large datastores on the Nimbles and allow DRS to control placement. WOuld best practices in this environment be to turn off DRS and manually place VM's on datastores configured for the same backup job?

[DGrinev]

Hi,

I'd recommend you to group VMs by type/customer to separate datastores as they might have different requirements for RPO\RTO. In this case there will be one storage snapshot created at a time.

WOuld best practices in this environment be to turn off DRS and manually place VM's on datastores configured for the same backup job?

This sounds logical indeed. Thanks!

[thyda.phit@hpe.com]

Hi experts

https://helpcenter.veeam.com/archive/ba ... iw_hp.html
Veeam Backup & Replication reads and transports VM data blocks via the backup proxy to the backup repository. For incremental backup or replication, Veeam Backup & Replication uses obtained CBT data to retrieve only changed data blocks from the storage snapshot.

Regarding the above explanation,
For Backup from Primary Storage snapshot having repository stored in the same primary storage, does the changed data is transported and copied to the repository volume?
If so, how to calculate the size of repository volume?

[Andreas Neufert]

Hello,

there is no point in having a backup on the same storage system than you primary workload. This is seriously dangerous.
Can you please explain a bit more?

Backup from Storage Snapshot read frist time a full, compress and dedup it and transport it to the Repository server. This server store it in the backup target storage. These roles can be one server or multiple.

Calculation for Backup target storage can be found here: http://rps.dewin.me/

If the question is that you want to work with Snapshot only jobs and how many data the snapshots consume on the storage, then you need to analyse the change rate on the VMs and speak with HPE about the consumtion (depending on storage settings and storage model).

[thyda.phit@hpe.com]

Hi Andreas
Thanks for the reply.

Our proposal have 2 nimble storages as primary for production and secondary for replication.
The Veeam backup server(proxy/repository) is a physical machine connected to nimble by iscsi.
What is the best practice for Veeam backup in this case, in order to restore VM using the following features.
-Veeam Explorer for Storage Snapshots,Veeam Explorer for Microsoft Active Directory
-On-Demand Sandbox for Storage Snapshots

Still thinking about the merit between using Snapshot only jobs and Backup from Primary Storage snapshot.

[Andreas Neufert]

We always try to implement at minimum the 3-2-1 rule. https://www.veeam.com/blog/the-3-2-1-0- ... ility.html
In you case if the second storage system is on another site, then you need to implement a media break as well.
You want to use primary snapshots, then replicate them to secondary system. With this approach there is no media logic break. If you source is corrupt (on VMware level) then your snapshots will have the same issues.Therefore you have to create backups as well.

Potentially you can keep some snapshots on primary, some on secondary storage and backup as well to secondary storage to another area for long term backup usage.