Disaster recovery orchestration for the Enterprise (formerly Veeam Availability Orchestrator)
Post Reply
chrisflyckelen
Service Provider
Posts: 85
Liked: 12 times
Joined: Oct 15, 2019 7:51 am
Full Name: Christian van Eickelen
Contact:

Case# 07914741 - DataLabs tests with 9+ VLANs

Post by chrisflyckelen »

Hey everyone,

I’m running into an issue while executing a DataLabs test with around 90 VMs distributed across more than nine production VLANs. It’s not possible to attach all of them to a single proxy appliance due to a VMware vSphere limitation: the proxy appliance supports a maximum of nine additional vNICs.

The actual restore works, but VRO registers the restored VM on one of the three ESXi hosts at random. As a result, some VMs end up on hosts different from the one running the proxy appliance, which causes connectivity issues during the process.

Has anyone implemented a setup with two proxy appliances running on different hosts that can communicate with each other to get around this limitation?

Thanks,
Christian
Alec King
VP, Product Management
Posts: 1604
Liked: 436 times
Joined: Jan 01, 2006 1:01 am
Location: Prague, CZ
Contact:

Re: Case# 07914741 - DataLabs tests with 9+ VLANs

Post by Alec King »

Hello,

I consulted our resident network guru and he does not think this would work... of course you could start two appliances on the same network e.g. a dVS or VLAN, but we don't see how you could usefully configure that.
Certainly we have no documentation on how to get around the vSphere limitation of 10 vNICs per VM.

Are all the VMs in the plan on a single storage volume? If there is a single volume, then in VRO it's a single Group and therefore a single Plan, and there is no way to split it further to try and reduce network requirements.

VRO will attempt to test all VMs in a storage plan, however you can disable testing for specific VMs (by editing the plan to remove the Register VM step from them) so that only your desired VMs start in the test. That would at least allow you to test all required VMs in a suitable lab.

You can control the host - and therefore the lab - which will be used for testing by editing the storage recovery location to remove all hosts except the host which runs the desired virtual lab. After the test, edit the location again to add the other two hosts back, and it will be ready for production recovery.
You could also manage the hosts in the storage recovery location externally, by changing the vSphere Tags in vCenter, testing, then changing them back. However there would be some delay while VRO collected the tags (could be several hours until the embedded Veeam ONE data collection is triggered)

Unfortunately the above are manual solutions, although some steps could be accomplished programatically. VRO does have the ability to execute pre- and post- plan scripts, and they can execute only when a plan is in 'test mode' for example.

Hope that helps!
Alec King
Vice President, Product Management
Veeam Software
Post Reply

Who is online

Users browsing this forum: No registered users and 3 guests