With a lot of the projects I work on, I have to do the overall design, detailed configs and the bill of materials. It’s always a fine line between trying to spec out the best network I can, and not wasting the customer’s money.

It’s called “meeting the business requirements”. If it really doesn’t matter if a remote site is off the air for a few hours, then it’s really not ethical to insist that they need dual supervisor cards, redundant power supplies, multiple line cards and diversely routed WAN links (although I know more than a few salesmen who would).

A decent maintenance contract and some spares will suffice. On the other hand, if it’s a major data centre, then it would be negligent to price up something that you wouldn’t even install in a test lab in your garage (what--you don’t have a test lab in your garage? Shame on you!), just so you can ‘upsell’ it later to something that’s actually fit for purpose once your company has won the bid.

There are so many occasions, unfortunately, when the business requirements aren’t clear (See “Salesman-driven design”, for example), when we have to try and second-guess what is needed and just do the best we can in designing a network that we believe is what is needed, isn’t overkill, and will do exactly what is required—even if nobody seems to know what that is. And that’s where our problems lie.

Too many techies are too honest. Much as we want lots of shiny new toys to play with, we seem to have it beaten into our heads, after years of being told that IT isn’t the core business of our company, that we are an overhead. A necessary (or even unnecessary, in some cases) evil.

I guess if your company’s raison d’etre is manufacturing widgets, then every year is an uphill struggle to persuade the powers that be that, yes, you do need some money to keep the network going, otherwise the manufacturing plant won’t actually manufacture anything—or your buyers be able to place any orders.

So we try our best to keep the costs down as much as possible, thinking we’re doing everyone a favour, and showing our corporate responsibility. That’s what happened a couple of weeks ago on a project I was working on.

We knew (kind of) what the requirements were, though they were rather vague. We had a network design (roughly). We just needed to get to the specifics so that we could detail how much money would be required.

So we started with our ‘ideal’ network, and started shaving bits off. “They’ll never go for that level of redundancy” we said. “We can cut back on hardware for some of the smaller sites if we have hot spares on-site”. “It won’t be quite as flexible if we don’t put this in, but it’ll save quite a bit of cash, and it’ll be okay”.

So we compromised, knowing that if we asked for what we really wanted—and in fact needed if we didn’t want to have to re-visit the design in a couple of years—it’d never get approved.

We really thought we were being good corporate citizens.

The paperwork started its progress through the layers of management. It got to the department head. He took one look at it, left all the techy information untouched, and slashed ten per cent off the bottom line cost. “I know what you techies are like—you always overestimate. Bet you’ve got a test lab hidden in there somewhere. And I’m not paying for more network management servers”.

At the next level up, the same thing happened—and the next. When we finally got the budget approved, the amount was significantly less than we’d asked for even after we’d cut the design to the bone. But nobody had questioned the design or bill of materials—just told us to do the same thing with less money. Now we had to make our already-compromised network work with even less kit.

Next time, I’m going to ask for three of everything. By the time it all gets approved, there might be just about enough to build what I’d spec’d up in the first place.