Feature Request: VM Startup groups in SureBackup

[segfault]

We have an application group at our DR site that contains 10 VM's for a SharePoint setup that we want to periodically validate. It currently takes about 50+ minutes to power them all on.

Within this application group are sub-groups of the VM's can be started in parallel. It would be nice to have a feature where we can specify a "startup group" attribute for each VM inside of the application group. VM's in the same group are started at the same time. This would drastically cut down on the time it takes to start the lab.

Using a linked job is not an option due to the lack of any start order with that method and the fact that we setup the replication policies by frequency and not application ("1 hour," "6 hour," and "daily" vs "Exchange", "SharePoint, "Wordpress," "JDE," etc), so each job will have dozens of unrelated VM's.

[Dima P.]

Hello John.

Sounds like a good idea. I've noted feature request based on this post and will discuss it with the team. Thank you for the input!

[knut-weber]

Hello, we have an Exchange Application Group with some Exchange Servers and Domain Controllers. About 15 VMs.
We can only adjust the start order. But we would like to start ALL VMs simultaneously becauce they have no dependencies and the overall start takes a very long time (about 2 hours) - although we are running copmpletely from storage snapshots.
Lowering the wait time for boot from 600 sec to 180 sec leads to failing VMs and Application Group stop.

Any ideas here?

[foggy]

Hi Knut, consider using linked jobs instead of the application group, there you can configure the number of simultaneously processed VMs.

[knut-weber]

That ist not possible to use linked jobs - and it makes no sense.
1.: we want to keep the VMs running for days because it is a TESTLAB for us.
2.: The linked jobs still need an application group
3.: Alle the resources come from different Backup jobs. All VMs from the jobs would be started, that would be dozens of servers.
4.: Some mashines are in NO backup job at all. These are DAG-members from Exchange coming from Pure Storage Snapshots only because they are not backed up. (We only backup Exchange systems with passive only DB copies)

We need no linked jobs, we need to start the defined resources in the group in parallel.

[zadrian]

I think by using linked jobs, it means the VMs will be powered on at the same time.
Then if the VMs are in no backup job, then is surebackup the right tool for the job ?

I would think that surebackup & surereplica are tools to verify the backups and/or replica respectively....this may not be the right tool to clone VMs into test labs ?

[knut-weber]

No, it is the perfect tool.
Linked jobs are stopped after testing - i want them keep on running. Application Groups do this. Replication would take endless hours.
It is OK to wait 2 hours until 15 VMs are up and running with 20 TB data in Exchange stores. It works great!

But there is no dependency, so there is no reason to wait for sequential start.
It could be running in 10 Minutes alltogether. That is the point.

[zadrian]

I do not get what you mean....
surebackup & surereplica is determined by all 3....the surebackup job that uses application group to determine the VMs to run on virtual labs. At the job properties, tick "keep VMs alive...."
With multiple application groups and virtual labs, I can boot up as many VMs as I need at the same time (as long I have the resources on ESXi or storage).
The link is on the running start of the surebackup jobs and not the shutting down....

[knut-weber]

I tell the AG to keep running. And the linked job is NONE. So I just start the job with no linked one and the application group.
I have several labs, several groups and several jobs.

It works all as desired! Evetything is OK. Our complete landscape is running fine inside. I can even send mail inside our lab to collegues working on a terminal server inside.

I just want to start *all* VMs in the application group in *parallel*. Thats the *only* thing I want. Everything else is great and working. I want 15 VMs be running in 10 minutes. I do not want to wait one after the other before starting. It would be just NICE to be able to select start-dependecies, dependencies in groups, sequential startup, parallel startup, ... ... ... . And not only move up / move down in sequential order.

[ChrisGundry]

I asked the same thing and was told it was done sequentially for several reasons:
1. If VM2 fails then there is no point in wasting time/resources booting VM3+ because VM2 failed as the overall job would fail because VM2 failed.
2. Powering up all VMs at the same time puts more load on the backup storage

I know neither of these issues apply in your case, but I think both are legacy issues from when labs was purely for SureBackup backup verification. Now that Veeam are pushing this as a virtual lab setup and have added options to run labs from storage snapshots and replicas I think they need to give us more options, like being able to power everything on at the same time.

Personally at the moment if I want to power up a larger lab and minimize the time it takes I turn off all the validation settings, all of the heartbeat, ping checks etc. This gets the lab up as quickly as possible and there is less time waiting between each VM booting.

[knut-weber]

Our PureStorage has some several 100000 I/O. That´s no point here. We use a dedicated ESX host. Load is no problem.
But indeed we unselect everything other startup options: VM heartbeat present.

I think not much people use it as a test lab. You need storage with snapshot providers to do this with acceptable overhead, so than no real disk blocks are transfered.
Of course it is cool to power up an Exchange DAG with 6 VMs, 24 CPU and 96 GB RAM each with 80 TB storage in 2 hours from backup - even if you have only backed up 2 of them with veeam and 4 are crash-consistent snapshots only. And everything runs fine! But when you *know* it should work in 10 minutes, you don´t want to wait longer.

[foggy]

Ok, guys, thanks for the details. I see the use case. Let's take this as a feature request then.

[segfault]

Side note, I made this request almost exactly a year ago:
veeam-backup-replication-f2/feature-req ... 54995.html

Same reason: once the AD controller is up, we can bulk start everything else. No need to go slow, we are using replicas at our DR site and the array is dedicated to the task and has zero impact on production. A lab that currently takes 2-3 hours to boot would be done in 10 minutes or so.

[foggy]

Thanks, John, I've merged the threads for easier tracking of the number of requests.

[ChrisGundry]

Out use case is similar, we use storage snapshots and have plenty of spare resource on storage and compute, so powering up a load of lab VMs at the same time would be no issue for us, but we have to wait for each to boot to heartbeat in turn before the next can boot, really slows the process down.