One armed loadbalancer placement across Multisite deployment

[GyroDom]

When using a one armed loadbalancer in a multisite scenario, i'm just thinking of the best way ensure high availability if one site went down.

Is there a way of splitting the loadbalancer HA Pair across datacentres ? or will i need to rely on SRM to fail it over in a DR scenario ?

Cheers

[GyroDom]

Some more thoughts on this one.

In my lab i've played around with this. I deployed an HA One armed loadbalancer in the primary DC.

Once deployed i migrated one of pair to the second DC.

The sync network between them was on a universal switch, so in theory would be still L2 adjacent, however when operating in separate datacentres the sync network stopped responding.

No further config could be pushed the cluster.

Due to the fact that the load balancers are managed by the primary NSX manager, i doubt SRM would work.

So based on this, i would expect the best mechanism for failover would be to create a second pair of load balancers and to manually configure them for each vip.

The second datacentre LB would need to have there NIC disabled so they don't conflict.

Happy to hear thoughts on this.

Cheers

[Richard__R]

My understanding is that the Edge appliances are managed by a VMCI channel so that would be a site-local thing which is probably why you can't split the halves of the HA pair out like that. Pretty sure that , as you say, keeping a redundant pair of edges on the secondary site that you sync manually is the only way that you can achieve what you want to do, they can't be failed over with SRM.