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

Designing CVP For Extremely Busy Contact Centers

[toc:faq]

Introduction

This document contains a design approach that one can possibly use to implement a CVP based extremely busy (High CPS) Contact Center Solution. This document discuss a unique scenario that could be faced by a very large Contact Center Solutions and it might not applicable to your environment. Also note that this is not a supported or tested design yet. Please work with your local Cisco SE or accounts team before implementing this solution.

Problem Description and Unique Scenario

Consider a hypothetical situation where there are calls coming into the contact center on the order of 200+ CPS, and 70% of the calls are self service calls. If all these calls go to the ICM, this would have exceeded ICM's capacity had we used a "normal" integration of ICM & CVP for all self-service & call center calls. Because the maximum capacity of ICM with a CVP standalone with ICM lookup is ~100 cps.

Proposed Solution Summary

What customers can do is to use the CVP Standalone VXML Self Service Server with ICM lookup model to front-end all of the calls, and then only calls that need call center treatment will get transferred to a second CVP farm that will integrate with ICM with queuing functionality.

Components

  • The first CVP Server farm (CVP-FARM1) is CVP Standalone VXML Self Service Server with ICM lookup
  • The second CVP Server farm (CVP-FARM2) is a comprehensive model with ICM
  •  Caller--PSTN-->Ingress-GW--CVP-FARM1--->CVP-FARM2--->ICM-->CUCM-->AgentIPPhone

Details

  • The Standalone CVP VXML (CVP-FARM1) does not have any signaling protocol, just VXML from the ingress gateway. When you do the blind transfer over to CVP comprehensive (CVP-FARM2) , then it will be H.323.
  • For CVP Standalone with ICM lookup, only those calls which require Call Center resources will result in a route-request to ICM. So only a portion of the incoming 200+ CPS to self-service will hit the ICM. That is the primary reason we have to split the CVP into two farms.
  • So majority of the incoming calls (70%) will be handled by the CVP Standalone VXML (CVP-FARM1)
  • CVP Standalone VXML (CVP-FARM1) will do a direct lookup to ICM only when the caller needs to talk to an agent. ICM lookup will also make sure that all the enterprise data collected by CVP standalone VXML (CVP-FARM1) is preserved too.
  • Since CVP-FARM1 is now a routing-client, the ICM will return a label back to CVP-FARM1.
  • CVP Standalone VXML (CVP-FARM1) will then blind transfer the call to CVP Comprehensive (CVP-FARM2) which will take the control of the call now. CVP-FARM1 is now out of the picture and call is queued at the CVP-FARM2 for further call treatment.
  • Notice that there is no queuing on the standalone. It is a self-service IVR only. When the caller opts to talk to an agent, the standalone farm    (CVP-FARM1) will do the route-request to ICM in order to pass the caller's data to ICM. ICM will immediately send a label back to CVP-FARM1. CVP-FARM1 will blind transfer the call over to CVP Comprehensive (CVP-FARM2). At that point, if an agent is available, CVP-FARM2 will transfer the caller to the agent. If not, then the caller queues on CVP-FARM2.

  • During all the time when the call was in the queue of CVP comprehensive, the call will be queued at the edge because the initial session was just VXML session.
Version history
Revision #:
1 of 1
Last update:
‎06-09-2009 04:37 AM
Updated by:
 
Labels (1)
Everyone's tags (3)