Tonight I went to the CouchDB meetup in Amsterdam (‘In De Wildeman’) to discuss the architecture I’m designing for upcoming VPRO projects (more on that in a following blogpost). We had a really nice discussion about mostly the ‘edges’ of what CouchDB can do. Impressive numbers, that’s for sure! Big sites/companies are showing interest (craigslist, yahoo, myspace, facebook, BBC).
Some interesting remarks/statements Jan Lehnardt and J. Chris Anderson made included:
- They are running CouchDB instances with multiple millions of records without any problems, hardware is the limit
- Writing speed doesn’t degrade to much when doing huge amounts of updates (they where talking about 1000 writes per second) as long as batches have clustered id’s (for instance sequential)
- Next to JSON CouchDB can do content negotiation, the sky is the limit!
- The team is working very hard on replication/sharding etc.
- You don’t have to know any Erlang to use CouchDB
- The mustache is the new beard!
Ow, and Kudos to @breun for organizing it!