Your current filters are…
by Sarah Mei
Diaspora is the crowd-funded open-source decentralized social network built on Rails. Full buzzword compliance: on by default. We have many thousands of active users and they generate a lot of social data. But after nine months of full-time development with MongoDB as our primary storage engine, a few months ago we converted it all to MySQL.
Wait...what? Most people are going the other way, dropping Mongo into a project in place of MySQL or PostgreSQL. Plus, conventional wisdom says that social data is ill-suited to a traditional data store.
Come hear a story about a large-scale Rails project that tried it both ways. You'll see crisis and redemption, facts and figures, nerds, kittens, ponycorns, and, of course, the secret sauce.
Hecklers will be piped to /dev/null.
by Tim Connor
An exploration of how the DataMapper pattern used by MongoMapper works very well with key value stores, in general, and how exceptionally well it works in a Document store like Mongo exceptionally, versus it's less ideal match with a schema based document store like an RDBMS. Tim will be comparing benefits of how the ease of modeling your data in such a way, versus the iconic DB modeling of 3rd normal form, which brings up issues of composition over inheritance, which favors 3rd normal and how ruby's mixin system plays into that tension.
This all leads up to having a data model that is trivial to make excessively dynamic, and challenging to keep sane and consistent, in such a system. Methods of migration and data massage are discussed with Tim suggesting what has worked best on some real Mongo projects, but such an approach would work well with any kv store.
29th September to 1st October 2011