I figure the two worst problems with a database are corrupt data and an unavailable system. You can follow good coding and referential integrity tips to avoid much of the first problem, and you can implement a few other tips to help with the second.
First, you need to define the causes of downtime. The primary issues are hardware problems, out-of-band memory corruptions, and service packs and other maintenace.
For the hardware, only use servers as servers. Make sure you follow all of the manufacturer's instructions and monitor the lifetimes of the drives and so on. That part most of us have down pat.
The second thing you can do is to implement a cluster. A cluster allows you to do maintenance on one system (including service packs) while the other is still running.
And don't forget to use filegroups. Using filegroups lets you stagger your maintenance as well.
Finally - upgrade. Many people are still running on SQL Server 2000, which runs a lot of processes (such as it's mail system) as an out-of-band process, which can cause the server to go unstable. Also, SQL Server 2005 has more enhancements to allow for real-time maintenance and so on.
Take advantage of special member promotions, everyday discounts, quick access to saved content, and more! Join Today.