Level Up: From Ops To Engineers

Talk given at DevOpsDays Austin 2013.

Level Up: From Ops To Engineers
Nick Silkey, Systems Automation Engineer, Rackspace

Gone are the days when sysadmins flung point-in-time commands at standalone, always up UNIX hulks. Now we are systems engineers whom are expected to treat their infrastructures as code, build fault-tolerant services, providing modular systems orchestration and ultimately deliver a turnkey software platforms to our developers. Whether you are just beginning this journey or are looking to level up, come quickly hear about foundations and applications of how to best treat your infrastructure as programmable, testable, deployable.

Transcript

1. level up : from ops to engineers devopsdays austin 2013
2. nick silkey @filler rackspace (we’re hiring)
3. traditional roles hand-rolled servers stable vs changing
4. releasing an afterthought tag = handoff = is it up yet?
5. failures => finger pointing => risk aversion => culture of no
6. if releasing is hard, you will delay releasing
7. get product to market quickly and accurately thats biz value
8. infrastructure as code “programmable, testable, deployable”
9. reconstruct the biz “source, backups, resources”
10. don’t manage machines, manage configurations version control > by hand
11. “cattle, not puppies” “wheat, not house plants” self-documenting, converged = compliant
12. modularize your infrastructure change, experiment => biz potential
13. liberate your infrastructure give devs prod parity; secrets are managable
14. engineer your releases commons tools for testing, deploying => closes ops/dev gaps
15. {log,graph} all the things => feedback loop for devs but only alert on some of the things
16. test all the things sufficiently tested => accurate releases => trust
17. cant fiat change, so identify adopters dev champ, ops champ
18. identify systems, v1 automate one simple function well
19. identify systems, v2 rebuild one system well, over and over
20. “be the automator, not the automated”