Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 

Hybrid CVP Deployment Model Design


Hybrid CVP Deployment Model

Co-Res Branch But Dedicated Central for Ingress GWs and VXML GWs


  • @ Branches Co-Resident Ingress GW+VXML GW AND
  • @ Central Location Ingress GWs and VXML GWs are on separate routers
If we had Co-resident Ingress GW + VXML GW at  both the remote and central locations
then “send to originator” will fix this issue


Customer Use Scenario

A customer who currently has a UCCE with SIP CVP centralized, comprehensive deployment. They are adding few or more remote sites where calls will ingress at the co-located Voice+VXML GW and the requirement is to to keep calls on remote VXML gateway for IVR treatment.

But on the central site (HQ) they have a farm of VXML GWs and they are not co-residing with the Ingress Voice GW. And the requirement is to queue the call at the one of the VXML GWs in the farm using load-balancing via SIP Proxy Server for those HQ calls.


Problem Statment

Questions is how to configure “send to originator”. Because it is a global config and once you configure it, the CVP will send the network vru label back to the originating gateway. Which will work in case of branch. Because in branch VXML GW and Ingress GW are co-residing. But it will fail the HQ calls. Because in HQ site the VRU Label (that is being sent from ICM to queue the call at the VXML GW)should go to the farm of VXML Servers.




  • For your remote site, the same GW must be acting as the VXML GW and Ingress GW (if not, “send to originator” is not an option)
  • On the central site, if the GWs are also acting as VXML GW and Ingress GW, then that would be easy, we could just configure “send to originator” on all the CVP call servers. And this command will make sure that call always goes to the originating gateway.
  • It is probably easy to figure out anyways. But the CVP is intelligent enough to distinguish that the label returned from the ICM is VRU label DN or Agent label DN. So for agent label DN, CVP doesn't apply the "Send To Originate" rule or command.
  • SetSigDigit works for both h323 and sip



Solution # 1

You could have a dedicated CVP server for the remote sites. All calls coming into the branches will be sent to that CVP call server and “send to originator” will be configured only on this "branch" CVP call server.

This will make sure that the call is always queued at the branch co-resident VXML GW for IVR treatment or for VXML interactions.


Branch------>CVP1(For Branch Calls with Send To Originator Command)

HQ---------->CVP2 (Without Send to originator command)


Solution # 2

  • Configure multiple customers/Network VRU, but that’s a configuration, reporting overhead and also it is not recommended. If you are going to use this option anyways, then make sure you design it very well and have A2Q approve this design before you implement.
  • You have to configure almost everything multiple times for each customer (with icmsetup and icm config manager), and it might have implications on configuration, administration, maintenance, and reporting.



Solution # 3 SetSigDigit

This is the best and recommended solution.

Configure SetSigDigit everywhere. For the central site: In order to prevent sending the call to VXML GW in the remote site.

Design Considerations for Solution#3


  • In the past, SigDigit solution doesn’t take ringtone DN and errortone DN into account.
 But this limitation is fixed now the the fix of CSCsv74353.


  • Requires extra call routing and configuration to be added to Ingress Gateway (add translation-route on the voip dial-peer pointing to CVP on every branch)