[VRA] - 2 DC deployment

[YannMichalak]

Hi,

Actually we work on vRA implementation.

We have TWO datacenters with not the same IP's.

vRA Appliance

DC 1 :

- Two vRA appliances

- One F5 LB vRA appliance with IP 10.0.0.1

DC 2 :

- One vRA appliance

- One F5 LB vRA appliance with IP 10.1.0.1

F5 LB appliance in DC1 & DC2 integrate the 3 appliances vRA (Two for DC1 & One for DC2)

When dc1 is unreachable we will change the load balancing DNS entry.

vRA Infrastructure Web

DC 1 :

- Two vRA Infrastructure Web

- One F5 LB vRA Infrastructure Web with IP 10.0.0.2

DC 2 :

- One vRA Infrastructure Web

- One F5 LB vRA Infrastructure Web with IP 10.2.0.2

F5 LB vRA Infrastructure Web in DC1 & DC2 integrate the 3 vRA Infrastructure Web (Two for DC1 & One for DC2)

When dc1 is unreachable we will change the load balancing DNS entry.

vRA Infrastructure Web

DC 1 :

- Two vRA Infrastructure Manager

- One F5 LB vRA Infrastructure Manager with IP 10.0.0.3

DC 2 :

- One vRA Infrastructure Manager

- One F5 LB vRA Infrastructure Manager with IP 10.3.0.3

F5 LB vRA Infrastructure Manager in DC1 & DC2 integrate the 3 vRA Infrastructure Manager (Two for DC1 & One for DC2)

When dc1 is unreachable we will change the load balancing DNS entry.

DEM & Proxy Agent

DC 1 :

- Two Infrastructure DEM & Two Proxy Agent

DC 2 :

- One Infrastructure DEM & One Proxy Agent

The purpose of this architecture is for it to be ACTIVE / ACTIVE and to continue to access @VRA Portal when DC1 is unreachable.

We can't, we wan't to use vMware Recovery Manager.

Do you think this type of infrastructure is viable ?

Thx for your feedback.

[daphnissov · Accepted solution]

This is not going to be a viable design for you for a couple of reasons.

1. VMware does not support a multi-DC, distributed design of vRA and its components. Read the KB statement here and look at the reference architecture here.
2. Two have active-active, café appliances, you still need a witness and that witness would also be down at DC1.
3. Manager components automatically fail over in 7.3, but you have 3 which is not a configuration that will work.

So at this point in time in vRA's design, something like this isn't possible while still being fully supported and functional in the use case you mention. Using a recovery method at site 2 is your only solid option.

[YannMichalak]

Thx @daphnissov :smileycheck: