Yesterday Eyal Efroni (aka – Shmeff) and myself participated in the first DevOps conference in Israel. The event, sponsored by Microsoft, GigaSpaces, Wix and others, perfectly organized by Sharone Ziteman (@shar1z), and led by Nati Shalom (@natishalom), Uri Cohen (@uri1803) and Ran Tavory (@rantav), was really a great experience for us.
First and foremost, it was a great opportunity to meet people who are as passionate as we are about DevOps. Most of the presentations were very interesting and insightful, but the crown jewel was the after-party at Mike’s Place which was a great opportunity to get to know people more personally.
But enough with all the melodramatic superlatives. Here are my main take aways from the conference. Feel free to comment and add your own insights:
Culture and Infrastructure
The DevOps movement revolves around two main axes: the first is Culture - any shift towards a DevOps oriented company will require a strong cultural basis or shift. The barriers between Dev and Ops need to be torn apart, and that starts with accountability: “you build it, you run it”. But accountability is not enough. In the end, it’s all about people’s willingness to break the old boundaries and cooperate. Getting your CTO/CEO/C*O buy-in is critical. Trying to break the distinctions between the two groups from the bottom-up is near impossible.
The second axis is Infrastructure - “if you build it, they will come”. It’s hard for developers to get used to the idea that they are accountable for the systems they build. It’s even harder to do so if they don’t have the necessary tools to take responsibility. Give them easy to use tools or a robust infrastructure and they will more readily follow you into ops-land.
The important point here is that Culture and Infrastructure are complementary: Culture without Infrastructure will only make developers frustrated and tired, while the other way around might be a total waste of resources – spending weeks or even months on building an infrastructure that nobody uses, or intends to use, is pointess.
The current tools are way too fragmented
Chef, Glu, Puppet, CFEngine, Nagios, Munin, Logstash, Splunk, New Relic, AWS, Graphite and I could go on and on. And still many companies find themselves writing in-house tools to manage their operations! I honestly can’t blame them when even the “simple” task of selecting a deployment tool is not trivial (as Ran Tavori demonstrated in his lecture). The world of DevOps is a complex one, but that’s part of the fun – making sense of all that mess. I believe that BigPanda will take an important part in making this fun even more fun (yes… that was a teaser).
Automation is King
In ops-land automation is king. It enables companies to focus on what really matters – bringing more value to customers – and not on constantly preventing their castle of sand from falling apart. Automate whenever you can. There’s a huge community of both people and tools that can help you do that. Use them!
It’s not all about success
Admitting your failures and talking about them openly, as demonstrated by Clive Foley, is an important part of understanding what works and what doesn’t. It teaches us even more than another success story that most of us can’t relate to. I lost count of the number of people who personally came to Clive to thank him (including myself).
I’ll summarize this post by paraphrasing Ben Kepes’ (@benkepes) closing statement: “it’s all about the people”. And I’ll add – we are the people, so let’s face it folks – it’s all about us.