Operability.io - Day one

Today I’ve been attending operability.io, a conference about DevOps but focusing on the Ops part of the term. The conference has only one track and the speakers are hand-picked, seriously cool speakers chosen by the organization.

In general terms, I have found that the talks have been about philosophical or psychological aspects of DevOps and not about technical aspects. I get that probably this is the leitmotiv of this conference (principles over tools, and so), but code and demos definitely beat slides with pics and cool sentences :-)

I have taken some notes regarding things that I have found interesting:

  • Andrew Clay Shafer (@littleidea) has mentioned the Google’s research Borg Paper. Almost every task run under Borg contains a builtin HTTP server that publishes information about the health of the task and thousands of performance metrics. Food for thought!

  • Colin Humphreys (@hatofmonkeys) has told a hilarious story about Monopoly City Streets game, a.k.a. waiting for 20.000 evenly distributed users during 3 months and ending with 5 million users. The PHP “True” of “False” quiz was really funny.

  • Anthony Eden (@aeden) has spoken about processes in DNSimple. On my view, too many processes and written steps but I agree with him that having public status pages and public post-mortems is a good idea.

  • Bridget Kromhout (@bridgetkromhout) discussed about how to work on distributed teams. Good advice as:

    1. Durable communications are inclusive, transparent and reliable
    2. In distributed teams, overcommunicating is better than being hurt or angry for no reason due to a misunderstanding
  • Collin Hemmings (@thegonzohunter) told us about dashboards not only for management or ops monitoring but also for developers (pipeline visualization and Monitoring Driven Development)

  • Elik Eizenberg (@elik_eizenberg) spoke about the principles of alert correlation. A bit too much theoretical, from my point of view, but otherwise nice introduction to the subject.

  • Kief Morris (@kief) discussed about automation and agile. How Toyota was not searching for cost reduction when automating, but for reducing lead times. He has mentioned the antifragile concept that I did not know.

  • Anne Currie (@anne_e_currie) explained that containerization speed up time from the dev point of view and increases datacenter capacity efficiency from the ops perspective. She said that more intelligent container schedulers are the future.

More comming tomorrow!