Mozilla2 Unified Storage

Stefano Mazzocchi: They are clearly looking for a native triple store, but they are confusing it with a embedded relational database. Stefano spotted an important architectural decision being made by the Mozilla architects. I had already heard that Mozilla was moving away from the RDF format because it was too slow. However, their wiki page on the subject, shows a much bigger problem they have to deal with in regards to storage formats. Stefano suggests Redland and indeed this is a very good candidate. From the top of my head, I think some of the reasons are:
  • Dave Beckett already did most of the hard work to make Redland cross-platform.
  • Mozilla developers already know RDF.
  • No need for db upgrade scripts on every Mozilla release.
  • No need for db install/upgrade scripts for Mozilla plugins.
  • Not exactly what Bosworth asks, but it’d be a “flexible” database.
One thing the Mozilla developers would have to work on would be Named Graphs. They might already be using this concept to manage all of their RDF models, but it would be extremely important in order to make querying more efficient and to have a nice mapping to SPARQL. Anyways, I wonder which RDF zealot will be proselytizing to the Mozilla developers about Redland. Many times we find ourselves arguing XML vs. RDF, but I have always thought that the real argument, technologically speaking is Relational vs. RDF. I’m nobody to say that relational dbms are broken, suck or don’t scale, but data can no longer be contained inside a single UML model.

About this entry

You’re currently reading “Mozilla2 Unified Storage,” an entry on Elias Torres

Published:
08.14.05 / 7pm
Category:
SemWeb