I know one day late, but here we go with my notes on DevOpsDays Day 2!. As it usually happens with these two days conferences, second day was a bit lower than first. Many people leave the same day (including me) and many others are sleepy (including me) from the previous day party :)
So here we go with a few notes:
Running a fully self-organizing/self-managing team (or company)
David Zwieback did a theoretical essay on how to better lead teams and how employee engagement is absolutely important for success. And engaged employee will align better with company values, will provide customers a better support, will try harder to bring value in any way she can.
The thing is how to create the best set of conditions for this engagement. One to ones, focused on digging employees strengths and the things that make them tick paired with very high internal flexibility to allow employees to drift into areas they are more fitted/attracted, were discussed as a good path to it.
Absolute transparency (on revenue, customers, projects, targets ...) was seen also as a key enabler to this employee self-selection of the areas she is best fitted for.
Unluckily, I didn't have much time to attend open spaces before leaving. But I joined the Immutable Servers one, where we tried to find out what the concept was about, how we could better apply it and which were the best use cases.
No one had much experience on it, but the common agreement was that the concept linked best with that of micro-services. Even if some people said that immutable servers don't really need to be linked to docker (guess not be caught following the hype), the conversation inevitably lead to it.
We discussed a workflow in which you split your application in the different services it is composed of. Then you deploy each of these services on its own (i.e.) container. Common agreement was just to use this approach for the web layer, that it is where it has more sense. And then you just treat each of these containers as an immutable server. No changes. You keep repository of images, and you updates these images when you need to do any change to a service (i.e. you could have an automated image builder based on some configuration management tools that will upload a new image to the repository at will).
If you need to update the service, then you just drop the running container and replace it with and updated version from the repository. You have a separate development pipeline for each of the services and in the end, you have a quite flexible and easy to update infrastructure.
Now it is just a matter of trying!
As usual, DevOpsDays has provided for me an extraordinary opportunity of getting out of my own small world and having a glimpse of what other people is doing, how they are solving their problems, which are their experiences and above all, listen a lot and absorb as much as possible.
What is great about DevOpsDays is that each one is an absolutely different experience entirely based on the random group of people gathered and their mood. What I have got from my two visits is much more than I ever anticipated, so check the one closest to you and give it a try!