Fail over In VCS cluster with DNS Round-Robin design

[Arbenjamin]

Dear all,

In the case of peer failed in VCS cluster, we would like to know the working priniciple of endpoint re-registeration.

The video endpoint will be re-registrated to other VCS peer by inqury the DNS server or not?

Or the Video endpoint gather the all VCS peers and saved in the DNS record from the first DNS inquery.

In the Cisco VCS Administrator Guide (X7.2), page 154.

SIP registrations

You can also use DNS round-robin techniques to implement a registration failover strategy. Some SIP UAs,

such as Movi™ v2.0 (or later) clients, can be configured with a SIP server address that is an FQDN. If the

FQDN resolves to a round-robin DNS record populated with the IP addresses of all the peers in the cluster,

then this could allow the endpoint to re-register with another peer if its connection to the original peer is lost.

Best Regards,

Ben

[ahmashar]

HI Ben,

if Movi/Jabber/endpoint loses its registration, it would inquire DNS again and DNS based on configured algorithm (e.g. round-robin) will send the next IP address of the peer in the list at where endpoint tries to re-register again.

regards,Ahmad

[Arbenjamin]

Dear Ahmad,

Thank for your kindly support.

We have understanded that video endpoint device (Movi/Jabber/endpoint) will inquire the DNS, however, how can to prevent the video endpoint device re-register to failed VCS?  As we would like to ensure the minimum downtime.

Best Regards,

Ben

[Martin Koch]

Round robin is not the ideal method for a failover.

Check out the vcs guides (like the admin or the vcs cluster deployment guide), taken from that:

Appendix 5 – Configuring endpoints to work with a VCS cluster...................................54 
H.323 endpoints ....................................................................................................................................54 
Option 1 – DNS SRV (preferred).................................................................................................... 54 
Option 2 – DNS Round-Robin (2nd choice) ................................................................................... 54 
Option 3 – Static IP (least preferred).............................................................................................. 55 
SIP endpoints ........................................................................................................................................ 55 
Option 1 – SIP Outbound (preferred) ............................................................................................. 55 
Option 2 – DNS SRV (2nd choice) .................................................................................................. 56 
Option 3 – DNS Round-Robin (3rd choice) ..................................................................................... 56 
Option 4 – Static 
IP (least preferred).............................................................................................. 56 

You should check out SIP Outbound and SRV if you want a better fail over.

[ahmashar · Accepted solution]

sorry forgot to reply to you. as you can see from Martin's post, there are different method of configuring DNS. on the endpoint you should set outbound parameter to "on" for a cluster of VCSs.

[Arbenjamin]

Dear Ahmad,

Thank for your kindly support.

I have checked the administration guide,

On: Set up multiple outbound connections to servers in the Proxy Address list.

In this case, the endpoint can register to other VCS peer whenever the first VCS connection is lost?

Best Regards,

Ben    

[Darin Walker]

I suggest you look into SIP outbound for physical endpoints. I just had this deployed two weeks ago on our VCS cluster. Endpoint are actively registered to however many VCS servers you wish. You can verify in each endpoint in VCS if the SIP outbound is functioning properly.

Sent from Cisco Technical Support iPhone App

[Arbenjamin]

Dear all,

We have formed with cluster with both methods(Round-Robin & SRV) by using the same DNS server,

C-series and MXP endpoint work fine in Round - Robin ,

however, MXP work fine in SRV record while C-series failed to registered to cluster name.(it can register to FQDN of vcs master and slave).

Anyone have such expierence?

Does C-series endpoint support SRV record ?

Finally, We only depoly H323 only in this testing environment.

Best Regards,

ben