vROps DR architecture design

[Peter_Grant]

I'm designing vRops for an environment that has a production environment and a DR environment. The customer has a general rule to provide DR between Prod -> DR and I'm considering the best way to do this from a vROps perspective.

Key Points:

1. Prod and DR is similar hardware (Vblocks)

2. SRM is deployed and available

Options:

Opt 1. Use SRM to failover the nodes and bring up the same vROps instance on DC2 (Is it possible or supported to rename the nodes and/or re-ip if we needed to using SRM? I know some VMware appliances don't like this. )

Considerations: Does it make sense to try and seamlessly failover vRops to the DR datacenter, because even though the hardware, software components will be almost identical it is still physically separate kit. So for instance, an historical hardware alert from DC1 wont be applicable to DC2. There are a number of adapters including one for UCS, vCM, vCD, Hyperic, vCNS

Opt 2. Deploy two separate and independent instances of vROps, one monitoring DC1 an one monitoring DC2 - don't fail anything over if DC1 goes down.

Considerations: We'll have no historical alerts or performance baselines, but are they valid as its different hardware? Admittedly, the software services will be failed over and they will be the same but does it makes sense to monitor them as if they were still running on DC1

Opt 3??

I'm interested in the community thoughts as to ways to deploy vROps across this Active / Passive DC configuration

Thanks

[mark_j]

Depends on your scale of the data sources.  If it's realistic scale/resource-wise, deploy 2x instance, each monitoring both prod and DR workload/data sources. Use prod primarily, and DR if prod take a nap. I wouldn't suggest re-IP'ing vR Ops as part of your DR design.. have your infrastructure already stood up and running. You'll have more important things to worry about if you're in DR crisis.. And in case you were thinking about it.. DO NOT EVEN FATHOM DOING A STRETCH VROPS CLUSTER ACROSS PROD AND DR DATACENTERS.

[Windspirit]

First of all I would ask if vROPS is really needed in DR. In a DR scenario what is the real use of vROPS? What should be done.

Second, if you want vROPS in DR you need to understand the following:

vROPS measures vCenter objects, each object is a moref (managed object reference) that is unique to the vCenter. The VM in your protected VC is not the same as the failed over one. So if you fail over the objects that vCOPS will be monitoring will be the place-holder VMs that were created when you protected the VM. So the only think that vROPS can do for you in DR is to see the CURRENT details, forecasts, trends, KPIs etc will only come into itself after at least one week (more like 3-4). If you are 3 weeks in DR...something went seriously wrong.

So if you want to measure and view the current trends then I would suggest 2 vROPS installations.

[Peter_Grant]

Thanks both for your reply - that aligns with my thinking too