vSphere Replication: Deployment practices

[karl495]

Please bare with me as this is my first time to post a question and probably deploying vsphere replication.

Scenario:

- I have an HA setup with 3 host + 1 storage on the production 

- 1 DR Host with large capacity equal with the prod (currently in the same location)

- Both Prod and DR managed by single vCenter

Asking for your insights with the following questions:

1. Is it recommended to use 1 vcenter considering distance of the actual DR would be around 260KM or 162 Miles?

- are there recommended / maximum latency and bandwidth between the VC and ESXI host?

2. If I'll proceed with 1 VC, where do you recommend installing the replication appliance? 

- should the vSphere Replication Mgt Server (VRMS) located on the Prod or DR?

- should i install additional vsphere replication server to other sites?

[vbrowncoat]

1. vSphere Replication is dependent on vCenter so if you are concerned about recoverability you may want to think about how you will ensure the availability of VC (or think about having a second one at your recovery site)

2. The traffic flow for VR goes like this: source host (where replicated VM is running) > VR appliance > storage at target site (through host at target site). Given that, it makes the most sense to place the VR appliance (can be either VRMS or VRS) close to the host/storage at the target.

[karl495]

Cool! Thank you for your insights

I would also like to ask about SRM, does it have automatic failure detection to trigger a failover to DR?

[vbrowncoat]

Next time I recommend a new thread for a new question. 

Automatic failure detection - yes, it can alert if connection to the other SRM server is lost. It also supports running a recovery plan via API or vRO workflow. I wouldn't recommend doing this though. See: https://blogs.vmware.com/vsphere/2014/05/automate-failover-with-srm.html