How to Beat the CAP Theorem

[jducoeur]

Several weeks ago, my co-architect Tonytip sent around a link to this article on database programming; I've gradually worked my way through it as I've had time.

It's well worth reading for any serious programmer who is working with data, especially Big Data. It's not trivial going -- it breaks down the concept of data access right to its fundamentals, arguing that if you want true scalability, you really want to rethink how you approach your data. But it's very cogent stuff.

It is *especially* worthwhile for anybody who has been tempted to dismiss the NoSQL movement as a fad. He argues for a hybrid approach that has good characteristics for many problems, and scales quite naturally. I don't think it's a panacea -- by its hybrid nature, there's an element of duplication currently present that bothers me -- but I think he's moving in some very useful directions...

Comments

[zachkessin.livejournal.com]

Thanks. I'm finding that CAP applies in all sorts of weird places when you get to distributed systems, not just databases as normally shown. I'm pretty sure that it is one of the reasons why it would be impossible to have strong types in a language like Erlang.