Virtual Lab Questions

[mcsmithSOP]

We're running 30 VMs on three hosts. We're using fully automated DRS to balance the load across the hosts, but have enough capacity to run all the VMs on any two.

I'm in the process of setting up Application Groups and Virtual Labs so we can set up SureBackup (one of the main reasons we went with Enterprise over Standard) and I'm unsure about a couple things.

First, if I understand the documentation correctly, it sounds like a VL needs to be tied to specific host? To fully test one of my groups (using scripts) I'd have to power on six VMs and depending on how DRS has distributed the production VMs, it could be risky to put them all on one host. (And even then, which host is currently the best choice may not be in the future.)

Having had first hand experience manually over-allocating a host (really bad things happen), I'm not keen to repeat the experience. What happens if powering on a VL VM would over-allocate a host? Would the VL VM get moved to a different host (if so, how?) or would the automated DRS move one or more production VMs to a different host?

Second, due to various constraints beyond my control, not all of my backup jobs were/can be set up to map to a logical Application Group. Almost every Windows OS SureBackup job is going to require the DNS and DC VMs, but in some cases I won't be able to link to a specific backup job so I'll have to select from Infrastructure. How do I exclude the DNS and DC VMs from being repeatedly verified in different SureBackup jobs? (We're using BIND for advanced DNS configuration and transferring the zone to the DC's DNS.)

Third, one of my VMs (SAP Crystal Server) is set to run a number of scheduled jobs to either a physical file server or email. Obviously, if I segregate that lab so it doesn't have access to the physical server or the Internet it won't be able to complete the jobs. My expectation is that it will generate a ton of errors which will "disappear" when it's powered off. However, in the future I'm going to need to use the VL to test an upgrade install which will need access to the internet and/or access to (some) of the production network. Any ideas on how to prevent that? My assumption is that as soon as the backup VM is powered on, it's going to think that it missed a bunch of scheduled jobs and start processing them before I can get in to stop them.

[PTide]

Hi,

like VL needs to be tied to specific host?

Yes, but there is a workaround provided by one of our engineers. Please see this blogpost.

What happens if powering on a VL VM would over-allocate a host?

I really doubt that a VL can over-allocate your host. In fact when your SureBackup job starts it first spins up your VL that is a very tiny VM. After that SB job starts VMs that you want to be tested one by one so it's more likely that one of those VMs will be moved to a different host.

I'll have to select from Infrastructure

When you choose "from Infrastructure" SureBackup job browses for VMs so that if you pick a VM then SureBackup job will try to find a backup for that VM. In this case you can choose any VMs but if none of them has backup available then SureBackup job will fail.

I'm going to need to use the VL to test an upgrade install which will need access to the internet and/or access to (some) of the production network.

For internet access from Virtual Lab please check this article.

Any ideas on how to prevent that? My assumption is that as soon as the backup VM is powered on, it's going to think that it missed a bunch of scheduled jobs and start processing them

I suggest you to make a replica of your SAP VM and disable scheduled jobs on that replica.

[mcsmithSOP]

Thank you very much. After several hours, several dozen views and no replies, I put it through Support and your answers are way better than theirs.

[PTide]

I put it through Support and your answers are way better than theirs.

Could you post your case ID please?

[mcsmithSOP]

Sure. Case # 01225798

To be honest, I was a little surprised because I've always had great responses from Support, but this one made me wonder if it was a former Veritas (Symantec) employee.

[PTide]

Thank you for case ID.

If you're not satisfied with the solution provided, you can always ask to escalate your case to the support management. All you need to do is to click "Talk to a Manager" button on the customer support portal.

Do any questions/issues remain unresolved at the moment?

[mcsmithSOP]

My main concern was that I know DRS manages the production load balancing fine, but I was concerned that if the VL VMs fired up too quickly, it would not give DRS enough time to move production VMs to a different host. Your reply indicates that is not an issue, so everything's good. Thanks.

[foggy]

Just a little note: while VMs in application group are processed sequentially, for linked jobs you can configure the number of simultaneously processed VMs.