Chat Ops
Because you're already in a chat room all day anyway
Jim Wrubel
CTO
Let's do a Julia Child thing
We're Think Through Math.
Annual Stats
3 million students in the U.S.
1 billion math problems solved
100 million hours of usage
8 Developers.
Automation is scale.
Communication constrains.
n(n-1)\over2
2-Pizza Teams
Dunbar's Number
Okay, enough with the the teaching
users here?
Any
February 2014
450 million users
35 employees
September 2015
900 million users
50 employees
Scale is impact.
Automation is scale.
Back to our chat command...
So what did this do?
Let's break this down
Hipchat: Slack, but for old people
Mathbot: Our hubot. Our ChatOps scripts are just node.js hubot plugins
Jenkins: Open source CI server
Scalr: API layer on top of AWS
AWS: Basically, the internet
Unicorn: Our primary app webservers (ruby-based)
TTM ChatOps Guided Tour
A subtle plug for Jeff Koenig
He's TTM's Manager of QA and the architect of most of what you have seen
(He's also a graduate of this program)
Questions
Feel free to use or remix these as you see fit.
Questions? Want to connect?
- Github: https://github.com/jwrubel
- twitter: https://twitter.com/jameswrubel
- LinkedIn (if that's still a thing people do): https://linkedin.com/in/jwrubel
ChatOps
By James Wrubel
ChatOps
Pitt CS Capstone presentation slides
