Thread by @swardley

X : Thoughts on DevOps?
Me : Broad subject.
X : General trends?
Me : DevOps / IaaS is what you should have been doing over a decade ago. It's now in the position of becoming the new legacy. The smart kids have already gone serverless and the practices emerging around it. Why?

X : We're just about to embark on implementing DevOps and cloud.
Me : Not unusual. It's ahead of many. Some are still waking up to the existence of the internet / www - see digital transformation.
X : So, it's a good thing to do?
Me : No, it's daft but not as daft as some.

X : Do you have examples of ...
Me : Are you looking for reasons to justify your inertia? This is no different with cloud (IaaS) in 2011. The desire to go down a path of virtual data centres because of pre-existing assets and capital? It's why people like kubernetes.

If your answer today is DevOps, IaaS, OpenStack, Private Cloud or Kubernetes then I would tend to say "sweat the existing assets for another five years" and then ask the question again.

X : Why are you so keen on serverless?
Me : Because it's boring. It takes all those things we needed to do to create a scaleable run time and just gives them to you ... pre shaved Yaks. The exciting stuff is what we build upon it and the new practices that emerge.

The things that change the world are the boring things. Maudslay made the nut and bolt boring with a screw cutting lathe hence kicking off the industrial age. Tesla and Westinghouse made electricity boring by making it a utility hence kicking off the age of electricity.

Amazon did this to computing. We went from shiny data centres with towers of servers, a sort of cathedral of computing with its own priesthood to buy as much you want for as little as you need on a credit card. They made computers so boring that we stopped naming them.

Serverless is doing the same to the runtime. That's where you should go.

Follow the boring, the money follows soon enough, then after much gnashing of teeth come the majority and finally (after 20-30 years of denial) the laggards appear claiming they were doing it all along.

X : I disagree. These "paradigm" shifts are never all encompassing.
Me : There are always niches but the majority (even laggards) are heading to cloud. Same will happen in serverless.
X : I disagree.
Me : That's ok. If the majority agreed with me then I would be concerned.

X : What do you mean by that?
Me : These "obvious" changes, caused by industrialisation, are only obvious to a few. Most have inertia, confirmation bias to past models and hence will dismiss the changes. If the majority agreed on serverless then I would be concerned.

Sitting in a room listening to a crowd of people tell me why serverless won't be massive in 2020 is exactly like sitting in a room listening to a crowd of people tell me why cloud won't be massive in 2011 is exactly like ... "Internet" in 1995 etc etc.

X : What about Google App Engine? It was first, before AWS lamba.
Me : Zimki was years before Google App Engine. Being first isn't the critical thing ask Sun, they were there before Amazon. Timing matters.

X : How does timing matter?
Me : Concept, suitability, technology and attitude (i.e. people hacked off with the existing paradigm) ... you need all those bits in place for people to play, for a community to grow and hence to trigger the change.

X : Any reason you can't mix DevOps and Serverless ?
Me : New factions create new flags. They often co-opt some of the past rituals but rarely admit to this. See DevOps and ITIL.