Friday, January 08, 2010

nosql

Accurate or no, the moniker "nosql" does seem to have legs in describing the growing interest in non-relational approaches to data management. Some interesting presentations here

3 comments:

Ben said...

By inaccurate are you talking about the fact that SQL != relational?

I'm working on a fully relational system that is also not SQL. (My efforts are pretty vaporous at the moment; even translating old notes over to something legible takes way more time than I expected.) And, it is annoying that fully relational stuff would get lumped in with the buzzword du jour crowd. Another example, the folks working on Rel shouldn't be lumped in as "nosql" either.

But the main thing that bugs me is that when you describe what you're doing as "not" something else, you're repudiating that something else. So nosql comes across as crapping all over the many talented folks who designed SQL, and that's not right.

To be clear, repudiation is not the same as criticism. Criticism implies actually learning something and understanding it and *then* pointing out where it doesn't work. But it's hard to do that with a buzzword like 'nosql.'

Reading the literature of most of these nosql products, it seems that they don't really know what SQL does, and the label itself indicates they don't understand the difference between SQL and the RM. SQL's strength is enforcing the everyday rules and processes that large businesses run on, so if your product has little or no integrity constraints, can't handle ad hoc queries, etc., it just can't do that. It might be a perfectly good technology, but it's not a competitor to SQL.

Sean said...

Ben,
Yes, by inaccurate I mean that SQL != relational.

The industry has mashed the two up to the point where they are terribly inter-twingled and fuzzy unfortunately.
Sean

Jay said...

The problem really is that it isn't NoSQL so much as NoMySQL, a reaction to that poor RDBMS and its limitations.