Perhaps the most critical of missing pieces to Amazon’s suite of web services launched yesterday. SimpleDB is Amazon’s database service.

Some people think that this is competing with enterprise database systems or the plethora of web-friendly databases such as MySQL. This isn’t really the case because SimpleDB differs in a number of respects:

• It’s schemaless, typeless and auto-indexed. This provides a more organic (some might think ‘ad-hoc’) architecture for storing and retrieving data. Everything is stored as UTF-8 strings.

• Each table (a ‘domain’ in SimpleDB terminology) is automatically replicated, but critically, consistency between copies can take seconds. Depending on the application, this is something that can be ignored, or needs to worked-around.

• The query language is very simple and queries that take longer than 5 seconds are likely to time-out.

• It is not possible to query across tables. This might appear to be a big problem, but whilst more traditional databases would require multiple tables to achieve an appropriate architecture, the typeless and schemaless nature of SimpleDB enables more to be done within one table.

• It is not possible to query and update in one operation. Instead you need to retrieve the result set, then perform the updates one by one. Similarly, there is no sorting of query results.

You might wonder after reading this, what good is SimpleDB? Well, if you need fast access to fuzzy data sets, SimpleDB would suit you fine.