cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
828
Views
10
Helpful
3
Replies

Internal service provision for Extension Mobility

James Hawkins
Level 8
Level 8

Surely this should be a no-brainer to enhance availability of Extension Mobility during a server failure.

I currently tend to use round-robin DNS with multiple A records to provide resilience but the 30-45 seconds delay if the failed server address is offered first is annoying.

Any ideas why EM cannot be an internally provisoned service (i.e. phone uses the CUCM with which it is registered rather than what is specified in a URL).

Any suggestions for other ways to improve this - would DNS SRV help??

3 Replies 3

Jonathan Schulenberg
Hall of Fame
Hall of Fame

That auto-node selection only works for services that have the Application:Cisco/CorporateDirectory format. Extension Mobility doesn't get this luxury for some reason. The only real way to do HA with it is with a load balancer.

Please remember to rate helpful responses and identify helpful or correct answers.

Ayodeji Okanlawon
VIP Alumni
VIP Alumni

James,

Like Jonathan has righly pointed out, the best/recommended way to provide resiliency is to use a load balancer.

You will need to provide resiliency for two things

1. EM service

2. Service directory URL.

A load balancer works on the concept of providing a virtual for use in configuring the EM service and the service url. When your phones request the EM service, the load balancer contacts the servers defined within the virtual IP. Any active server that responds then processes the request.

One side notet that is often overlooked is that for full resiliency, you need to have resiliency on your load balancer too, otherwise when that fails then you loose your EM function altogether

Please rate all useful posts

"'Nature is too thin a screen, the glory of the omnipresent God bursts through it everywhere"-Ralph Waldo Emerson

Please rate all useful posts

Thanks guys,

It is pretty odd that Cisco cannot improve this with CUCM particularly as they are exiting the load balancer market.

I guess one cheapish option would be to run software load balancers as VMs but it still seems a lot of effort to solve what the product should include by default (imo)