I have a vROPS 8, 2 data centers, and about 20 remote sites all with VMWare footprints. I wanted to configure vROPS to monitor the remote sites as well as within each data center and have DR coverage in the event DC1 goes down. I know i have to use continuous availability and a GSLB so that I can direct traffic to the active Master. I am only trying to protect from data center failure, and therefore I only want vROPS to failover to DC2 in the event DR is called in DC1. My question is around the witness node. Here is my thinking:
From the GSLB I point the WIP to each of the Master nodes.
I would build the vROPS master in DC1, then build the master replica node in DC2 and point everything to the WIP on the load balancer. Devices in DC1 would using the WIP point to whatever was the active node, the devices in DC2 using the remote collector point to whatever master node is active, and the remote sites using the WIP would point to whatever master node is active...
Is this the correct architecture, or am i missing something?
We had same requirements and after discussion with our VMware TAM the vROPs CI required 3rd site to work and cover all DR scenarios other wise you might encounter serious issues when you lose site with witness. You can find more details => https://docs.vmware.com/en/vRealize-Operations/8.6/com.vmware.vcom.refarch.doc/GUID-73E81793-74AF-46...
I can see 2 options :
- go back to HA cluster
- deploy witness node on VMware certified cloud provider (it is the cheapest option to have such setup if security will allow fort that 🙂 )
Hope that will help you.
I tested vROps CI in my test env without 3rd site for witness and that didn't go well when we lost one site and support was not able to recovery that cluster so it is clear for me that 3rd site for witness is a must
Hope that will help you make some good decisions for future.