Case Study: Why we rejected Voice over IP

I’m afraid that I am getting a little bit tired of being cold-called by salesmen wanting to arrange meetings and to take a ‘few details’. “What are your plans for convergence?” they ask. “What is your wireless strategy?” And so on.

I remember them asking much the same sort of question about ASPs a couple of years ago. On being told that we had no plans, or that our strategy was not to go anywhere near the technology in question, their tone of voice would usually imply that we were poor, unenlightened souls living in the dark ages. “May I ask why not? they’d say.

There is certainly a place for convergence. In the right circumstances, with the right topography and with the right business drivers, Voice Over IP (VoIP) makes sense… in the same way that in the right situation wireless LANs make sense… or outsourcing… or any one of a dozen other ‘must have’ technologies .

Well, in the case of VoIP, yes you can live in the dark ages quite happily – and here is why.

My employer, Eton College is a fairly large organisation; only eight to nine hundred staff in total, but occupying over four hundred buildings that make up a large part of the town. The staff live in flats and houses spread widely around the town. There are twenty five boys’ boarding houses, schoolrooms, libraries, sports facilities, the administration, plus all the usual facilities.

We have a conventional telephone system based on a PABX housed in the oldest part of the School, with telephones fed by copper pairs in most of our buildings. We also have an extensive data network based on a 622Mbps ATM core (remember ATM?) with switched Ethernet edge devices and dedicated outlets in every master and boy’s room and in many other locations.

The PABX is full. It does not provide us with all of the facilities we need, particularly in the areas of tracing and blocking nuisance calls. A project was initiated to find a suitable replacement system. If the prospect of such a procurement already has VOIP suppliers salivating in anticipation then the bad news is that we have already placed a contract – for a conventional solution.

We shortlisted three suppliers, Telewest with their Centrex offering, and NEC and Bistech both with VOIP solutions. Costs were broken down over a five year period, including the installation costs and the anticipated running costs based on current traffic levels. All three systems offered us savings over a five year period of around £120,000. The surprise was that they were all within a few thousand pounds of each other. The point is, of course, that topography and the nature of the legacy system really determine the extent of any savings. Though our buildings are spread around the town we do only have the one ‘site’. There are no opportunities for savings on ‘internal’ traffic.

The VOIP solutions would have required extensive re-cabling as the topography of the existing telephone network is quite different to that of the data network. This expense was, however, qualified by Telewest’s view that most of our existing copper cabling needed replacing anyway.

With costs roughly equal, the decision rested on technical and operational criteria. A case can always be made for new technology on the basis that it is ‘future-proofed’, though this is not a good strategy when selling a system to a man running an ATM network…! It is also possible to point to the wealth of additional features which, if not already available, are just around the corner – honest. We are but simple users in terms of telephony, and whereas some of these features look nice on paper, it is pretty hard to justify them when set against the drawbacks.

The main drawbacks are operational. I told the VOIP resellers that I was not confident that our data network could provide a reliable enough service for our telephone users. One of the suppliers offered to build an entirely separate network alongside our existing network, just to carry the telephony. They didn’t seem to quite grasp the nature of the problem.

Our central 622Mbps ring feeds switches in many other locations around the School - between fifty and sixty of them. Some of these are in inhospitable environments or cramped spaces. They do not all have uninterruptible power supplies. Naturally we manage this network to the best of our ability and put a great deal of effort into making it as reliable as possible. We protect the backbone, server rooms and other vital functions with as much power regulation as we can afford, but it makes no economic sense to do so for all of our edge devices.

Most of our network issues are related to power problems. Take the case of a boarding house, home to fifty boys and related staff. A power failure in the middle of the night can leave the house without network access for some hours. We don’t have twenty four hour cover for our data network – most people are asleep at night in any case. Even so, the phones in the boarding house must continue to operate – an absolute requirement for safety.

With a conventional exchange in a central location we can get a duty operator to investigate any problems; with the Centrex solution we don’t even have to do this ourselves. We simply cannot get access to most boarding house switches in the middle of the night and this in itself rules out the VOIP approach.

Conventional telephone systems are hugely reliable. Data networks are still less convincing. In a diverse data network it surely makes little sense loading additional services onto a large number of potentially vulnerable switches unless there is some real financial or operational gain. The criteria will always differ from case to case, whatever your vendor may wish you to believe. In our case the risks were just not worth taking.

This story, "Case Study: Why we rejected Voice over IP" was originally published by


Copyright © 2004 IDG Communications, Inc.

Shop Tech Products at Amazon