At NetWorld+Interop 2000 in Atlanta, voice over IP was everywhere. There was Voice Over IP Day, an impressive demonstration of voice-over-IP interoperability in the iLabs, and voice-over-IP vendors everywhere. Unfortunately, the voice-over-IP gurus are not so enthusiastic about quick absorption. At a postshow shindig, here's what I learned from the folks in the trenches.
In the toll bypass voice-over-IP area, the technology is ready even if the economics are not. Running voice over IP between two sites to avoid 4 cents per minute long-distance charges is a difficult proposition to cost-justify. You've spent $100,000 on a big voice-over-IP network, and you're already 2.5 million call minutes in the hole. Plus you've still got to build and pay for a QoS-based network to carry the call, using either quality of service to keep the voice running efficiently or quantity of service over-engineering. If the dollars work for you, then go ahead because voice-over-IP technology in the toll-bypass area works great.
Try to push voice over IP out to the edge in a company, to the phone on the desk, and you'll encounter four main obstacles:
The call managers and gatekeepers are not scaling to corporate levels. These are critical pieces. Try to present a serious call volume to existing gatekeepers, and you'll be looking at crash dumps and blue screens. Distributed call management helps somewhat, but when you put 5,000 or 10,000 phones on a corporate campus, the bottleneck is very real and the software is just not capable.
Voice over IP at the desk breaks Enhanced 911 location information. A voice-over-IP call comes from an IP address, not from a pair of copper wires carefully documented in a database. Where's that IP address? If you're lucky, you might be able to find the building. Any organization with thousands of phones will find the lack of integration with Enhanced 911 a showstopper.
Voice-over-IP quality at the edge just doesn't match what we're used to with a simple PBX. Quantization delays in the digital signal processors that convert analog to digital, along with inherent delays in IP packet building and processing, give a one-two punch to voice-over-IP quality. These phones have a noticeable, and annoying, delay. This doesn't signal a new life for ATM at the desk, but it does mean the voice-over-IP technology package needs to be fast to be acceptable.
There are the economic reasons. Voice-over-IP phones are dropping to the $400 to $500 range, but that's still substantially more than a PBX phone in the $100 to $200 range.
Does this mean voice over IP at the desk is dead? Absolutely not. The features voice-over-IP phones offer are outstanding, and corporate managers are running successful trials in smaller companies.
But don't budget for that 10,000-phone rollout until your vendor tells you how to solve these four problems.
This story, "The voice-over-IP push is premature" was originally published by Network World.