Tutorial: The All-IP Network
Lowering the Cost of Service Development

This is an archived article that appeared in the Teracom newsletter June 2004.

The Problem Carriers Need an Answer To

Operating costs, specifically, the cost of new service development. The rationale for the All-IP Network (formerly known as the all-ATM network) can be understood by understanding service development at a carrier.

To deploy a new service, a carrier has to undertake a mind-boggling number of steps and an enormous cost. For example, the carrier must:

• Develop a business case and detailed specification for the service.
• Design and deploy the network infrastructure for this service: the customer premise terminal and circuit-terminating equipment, the access technology or technologies between the customer and the Central Office (CO), the "edge" switches or routers that interface the access to the network core, the core routers or switches, and core transport capacity.
• File and have accepted a tariff, or an application for forbearance from regulation with all involved state and federal regulatory agencies.
• Make provision for the new service in the "business office" systems, for receiving customer orders, in the "network provisioning" systems for installation, and in the "repair bureau" systems for trouble and repairs. This would include amongst many other things, modifications to the multi-million-dollar mainframe-based billing systems and databases, and for repairs, selection and purchase of equipment for testing the service.
• Train the personnel in these three areas how to order, install and repair the new service.
• Develop a marketing and sales campaign. Update all the product brochures and web pages.

And a zillion other things. Many of the above items would require teams of between 5 and 20 people and person-years of labor.

Can you imagine how much this costs?

Probably not. The number of steps required may be beyond the ability of a single brain to comprehend.

This has to be undertaken for every new service, and every enhancement to every service.

For established carriers like Verizon, this means that the ability to deploy new services in any sort of rapid and cost effective manner is seriously impaired... but, they have revenue from existing services and existing customers to subsidize these costs and keep the company running while new services are being developed.

For startup carriers like XO Communications, the steps required and costs involved in deployment of new services may well be a business-ending hurdle, as they don't have existing customers to provide cash flow during service development. And even if they managed to develop an initial suite of services using seed money or IPO money, the requirement to deploy new faster, better, enhanced services every nine months might lead to Chapter 11.

We must keep in mind that this is a never-ending, and in fact, accelerating problem.

As the capacity of the physical transmission media grows exponentially via fiber optics and Dense Wave-Division Multiplexing, and the capacity of the end-users' data and media processing systems increases to unimagined levels of horsepower, the carriers will be continually pressed to introduce newer, faster and better services.

So how do we fix the problem?

That's easy. Just have one network service, that handles all types of communications, and is scalable to any access and transport line speed. One technology in the core of the network. One technology for access circuits. One brand name. One bill.

In the old days, this was to be SONET and ATM: Asynchronous Transfer Mode, where we would chop everything - voice, data, video - in to 48-byte chunks and carry them together over fiber, with various Quality of Service (QoS) guarantees for different applications, including Variable Bit Rate - Real Time service to carry voice in "packets" with guaranteed quality at the far end. But it never really happened.

Now, the answer is IP, where we package everything - voice, data, video - in IP packets and carry them together, with various QoS guarantees for different applications, including guaranteed quality of speech.

To do this, we will need supporting technologies and protocols: MPLS for IP QoS, Ethernet over copper, fiber and wireless for access, Optical Ethernet and Dense Wave Division Multiplexing on fiber for the physical transmission core.