Category — Common

Ten of the Biggest Mistakes Developers Make With Databases
By Mike Gunderloy
March 6, 2006

Although fashions come and go in software development, some things stay remarkably constant. One of these is the use of databases. You may be wonderfully up-to-date with an AJAX Web interface or the latest whizbang Windows user interface, but under the covers, you’re probably still pumping data in and out of a database, just as we all did a decade or more ago. That makes it all the more surprising that developers are still making the same database mistakes that date back to those good old days of Windows 95 and before. Perhaps it’s just that most of us learn to use databases on the side, rather than really studying them. In any case, here are my nominations for the biggest mistakes that I see over and over again.

Choosing the Wrong Database

Not all databases are created equal — which means before you do anything with a database, you have to pick the appropriate database. Time and again I’ve seen Access databases groaning to bear the load of huge data sets that would have been child’s play for SQL Server, or harried users trying to pay for and set up SQL Server to hold a few hundred rows of data. Broadly speaking, there are three tiers of databases in the market these days: desktop and embedded databases suitable for smaller tasks, “Express” versions of the major players that are good up to a few gigabytes of data, and the truly enterprise databases like SQL Server, Oracle, and DB2 that can handle just about anything you can throw at them. Before you do anything else, you need to make some realistic estimates about the amount of data that you’ll be storing and pick the appropriate product to do the storage.

Choosing Too Many Databases

APIs such as ODBC, JDBC, and OLE DB have promoted the notion of database independence - the idea that you can write your application code in such a manner that you can plug any database at all in for data storage. Well, yes, but there are compromises. I’ve seen a lot of teams go down the rat hole of database independence, writing layers to translate all of their SQL statements to some lowest common denominator dialect that every conceivable database will support, and at the same time giving up on advanced features available in any particular database. The notion seems to be that some client in the future might want to switch to Oracle or DB2 or FoxPro or whatever, so it’s best to be prepared now. On the contrary: when you’re starting out with a new product, pick your storage engine and write to it. If your product is good, people will install the database you specify, and you won’t be wasting untold man-hours supporting “just in case” scenarios that you’ll probably never need.

[Read more →]