Friendfeeds usage of mysql: schema-less

Interesting discussion on friendfeeds new database usage: How FriendFeed uses MySQL to store schema-less data. It’s basically a key-value database implemented on top of mysql, to get around schema-change and index performance problems in mysql.

There is a lot of activity around non-RDBMS databases recently. Although it’s always interesting to see new development and innovation, it’s not always clear if these new innovations are really necessary because RDBMSes can’t handle the task. Could it be that application-server developers just don’t know how to properly use a RDBMS, or maybe even refuse to use all that RDBMSes have to offer? Or maybe startup-developers are so hooked on gratis or free software that they refuse to use expensive, closed source commercial software?

The blog on friendfeed’s usage of MySQL again doesn’t answer the question if commercial RDBMSes like Oracle can’t handle the requirements. The post doesn’t mention if they also tested Oracle. Some comments argue that friendfeeds requirements shouldn’t be a problem for Oracle or SQL-server, others state the opposite. Whatever the truth, i fail to see how locking yourself into a middleware language is any better than locking yourself into a database language. Afterall, why not use a data oriented DSL for data related tasks?

blog comments powered by Disqus