All of us at Veriton believe that good database design is at the heart of a successful system. This doesn't necessarily mean 3rd Normal Form (or 4th or 5th NF for that matter), in fact more often than not it means careful denormalisation in places.

Veriton has a proven track record of designing successful databases to store millions of records. Our technique is based on sound theoretical understanding and practical experience. One of the principles we've found particularly helpful over the years is to model the data on the real-life situation facing the business as much as possible as it is this core structure at the heart of the business that is least likely to change.

For example if an electricity company has 2 million physical meters, bolted onto walls in homes, shops and offices, then there's a good chance that a METER entity describing each physical meter is not going to change very quickly! Certainly new meter types may be introduced that have new features, and therefore attributes and relationships, but this still won't negate the fact that there are millions of existing meters that aren't changing.

The difficult part if usually modelling the logical contructs: the "metering system", a "system configuration," a "tariff" - that's usually where the experience comes into the equation.

Over the years Veriton has developed core standards to cover the design and build of databases, covering naming conventions for entities and tables, attributes and columns, relationships and keys.

Our data-modelling tool of choice is Oracle's Designer 9i. However we also have ERwin and UML among other skills in our repertoire.