Performance benchmarking

[danielcompton]

trafficstars

It would be good to build performance benchmarks to catch any performance regressions.

[bhurlow]

maybe we could use docker to bootstrap a test db environment

[KeeganMyers]

I've been running all application services in their own docker instances for a while. It works extremely well so far. Just bind the required docker ports when setting up the instance

docker run -d -p 1090:8080 --name rethinkdb -v "$PWD:/data" rethinkdb

also if you run into issues binding the instance to a single ip this hack may help

(use [clojure.java.shell :only [sh]]) (:out (sh "sh" "-c" "docker inspect rethinkdb | grep IPAddress | awk {'print $2'}"))

[bhurlow]

@KeeganMyers nice! would be nice when considering performance to clearly define how the db is set up. Rethink allows for lots of configuration with caching, threading etc that could make for really different results

[danielcompton]

When I was thinking about this, I imagined we'd mock out the connection to RethinkDB to keep things reproducible, and to focus on the performance of the driver (which we control), and avoid any blips from the DB.

[bhurlow]

@danielcompton that's a great idea