MongoDB SF Conference Notes

CRUD in the JS shell
http://github.com/mdirolf/shell_presentation

Schema Design
4MB object limit
atomicity at the document level
... the rest of the talk isnt visible due to his slides/projector :(
the $ operator sounds cool, I need to research this
compare & swap is the safer and more appropriate pattern than just modifying a single value
Sharding should be considered when designing the schema
Capped Collection - a rrd(?) style. fixed size, will delete oldest records when the size limit is reached
    automatically stores insertion time and allows for queries based on that value

From Mysql to MongoDB
mongo loves system resources
    run on its own machine to keep from paging
    takes significantly more disk space than mysql
    disk speed is your bottleneck
mongo is faster than hibernate in java (woohoo?)
reduce disk usage by using shorter key names (veryLongAttributeName => vlan)

Mongomapper
custom types => DowncasedString (to_mongo/from_mongo)
gridfs + jnunemaker's joint plugin to store files
identity map plugin may help reduce queries but requires a rethink of how to use mongomapper
prophesying
    activemodel (when rails 3 is complete)
        validations, callbacks, dirty tracking, serialization, etc.
    blank document
        mongomapper w/o all the plugins (ie: more customization)
    mongo::query
        similar to ARel
Class in Michigan: ideafoundry.info/mongodb

Event Logging
map/reduce
counting in real time! (?)
use ruby to generate JS map/reduce code
mongo is really fast for map/reduce

Administration
log rotation commands built in

Rambling one post at a time