Apple hires...

There's a rule in the database business: the more useful a database is to an industry, the worse it will be written. This is because the people who know the most about a specific business aren't programmers; they're industry experts who have dabbled in a bit of programming.

These experts will pretty much always grab the wrong tool when making a database, and the expert knowledge which that database presents will be so vital that the users will, in the immortal words of Ken Olsen (founder of Digital Equipment Corp), "walk through fire to get to their data". We'd like to mention some howlers here, but our legal budget is slim...

There are some good databases out there - Circle, for instance, is the 300-pound gorilla in the property development and planning marketplace - but you won't hear about it outside its specialised markets. And, unfortunately, nor will you hear about the bad. The truth is that most operations with a vital and, hence, badly written database will move heaven and earth to support it.

One unexpected spin-off is that you should instinctively distrust computer support providers that continually whinge about them - too many support operations are trying to uphold an unattainable technical ideal, and the fact is there's always something like this in the mix, so they should just get on with it.

Some appalling databases add huge value to the businesses that run them, and the cost of running them or hiring an in-house support person should be factored into the equation. Theoretical ideals are attainable - but it's shocking how large the organisation has to be, and how simple the requirement, before they start to crop up. (Whatever you do, don't mention the National Health Service.)

Finally, a note of caution: WLANs and databases don't mix. If you can break this rule successfully, then it's because you have a high value, low volume data model - say, for instance, barcodes in a warehouse. Everyone else, use wires.