ucce deployment query

[ganeshraghuram]

Hi i am working on a ucce deployment which has side a at Florida and side b at virginia.We had recommended the customer to go with 2 diffrent subnets for the private and public communication as a best practise mentioned,however the customer has some limitations on giving the extra subnet.The customer is ready to take the responsibility of matching the required bandwidth and latency all the time between the private links and public links.So ideally it has to be a single nic for both private and public.They have an MPLS backbone with a redundancy in place.Would it be smart to go with a single subnet?I am definetly sceptical on that.Is there any other such networks running in this mode and if yes what are the problems which could arise?

[geoff]

As a Cisco partner, you will have to submit your design to Cisco as part of the A2Q (Assessment to quality) process.

There are now three A2Q reviews, the second of which is the most important. You have to pass this before Cisco will release your software order. The A2Q process is extensive, and the correct design of the private link between the sides is something that will be carefully examined. If you run public traffic over the same link as the private traffic, you will not pass A2Q.

This is not a question of subnets - but actual physical links. It can be difficult trying to convince the customer that the Cisco requirements are mandatory - but you will have much tougher time trying to convince Cisco to release software for a design that is deficient in this area. They are almost certain to reject your design.

The customer may feel that they can control the bandwidth, latency and QoS marking to meet the Cisco requirements, and with care and attention to detail, it can be made to work.

However, you open up the ICM system to a very serious situation should the link fail and the Call Routers lose the private link and the public link at the same time. Depending on how the PGs are arranged, you can get into a situation called "split brain", where both side A and side B independently think they are in control, so the Contact Center continues running. But when the private link comes back and the two Router/Loggers try to synchronize, they cannot. Now you are in serious trouble and the database may be irretrievably corrupt.

Regards,

Geoff

[Muhammad Raza]

Best Practices

•

•

Unified ICM Central Controllers (Routers and Loggers) should have an separate network path or link to carry the private communication between the two redundant sites. In a non-distributed Unified ICM model, the private traffic usually traverses an Ethernet crossover cable or LAN connected directly between the side A and side B Unified ICM Central Controller components. In the distributed Unified ICM model, the private communication between the A and B Unified ICM components should travel across a dedicated link with at least as much bandwidth as a T1 line.

The private link cannot traverse the same path as public traffic. The private link must have path diversity and must reside on a link that is completely path-independent from Unified ICM public traffic. This link is used as part of the system fault tolerant design.

We are handling a distributed ICM deployment since last 3 years, i would recommend to use two different physical links for visible and private.

Regards,

Muhammad AR

[ganeshraghuram]

Thanks Geoff and Amir.Well its quite clear then that i will have to tell the customer to give us separate subnets for private and visible.Passing the traffic over the same mpls link is something A2Q can give an approval for.