Thread by @swardley

X : Why do you think severless will beat containers?
Me : An irrelevant question. Containers might win the battle below the runtime but who cares? The future is the runtime and above, that's where the war is ...

... what will catch people out is the punctuated equilibrium nature of such a change i.e. its exponential growth. Same happened with EC2, five years in (2011) and people were still dismissing it as irrelevant.

As I said about cloud, so I will say about serverless ... this is not an 'if' question but a 'when'. There will be co-evolved practices, there will be a billion dollar company that just produces a function, there is no choice.

Serverless is also in that early phase of no-one caring too much about costs, as it tends to be vastly cheaper than alternatives ... but don't make the same mistake of thinking that serverless will reduce your IT budget. It won't. You'll just end up doing more stuff.

X : Serverless is more than the rutime.
Me : Your serverless architecture is a mix of the runtime and the services you consume.
X : Containers are part of that.
Me : They are irrelevant to a serverless world, a low level detail that you should not care about.

X : But you can't build a company on serverless?
Me : You can't build a company on EC2.
X : Yes, you can.
Me : Not according to lots of people in 2011. They were wrong. So are you.

X : Containers are more popular than serverless.
Me : Your point being? In 2010, DVDs were vastly more popular than streaming / digital alternatives.
X : That's different.
Me : True. DVD folk used to laugh more loudly over the idea that they were not the future.

X : Any recommendations on serverless?
Me : Beyond what I said last year ... no.

X : So you do think serverless will beat containers?
Me : It's an irrelevant question. They are not competing in the same space. Serverless is batting over the future, the runtime and above. Containers are battling over plumbing that most of us are not going to care about.

I.e. what you should be looking at is Riff vs Lambda vs Azure Functions vs Google Functions. The focus should be on the runtime and above i.e. not providing the runtime and certainly not below.

X : but serverless has evolved beyond function as a service.
Me : The runtime as a utility has evolved beyond the runtime as a utility? That's some fine marketing weed that you're smoking. Certainly we will be moving up the stack i.e. conversational programming.

X : I disagree.
Me : Well, time will tell. We will know 10-15 years after industrialisation started i.e. 2024 to 2029 is when Serverless (runtime as a utility) should become seen as the new "norm" for the majority. The laggards will take much longer. It's not long to wait.

X : Did you see that 12,000 people were at KubeCon.
Me : And?
X : 12,000 people disagree with you over the future of containers.
Me : Hundreds of thousands of people disagreed with me over the future of cloud. Your point? 12,000 can't be wrong?

X : You don't see a future for containers?
Me : I don't see it as relevant in the long term. Sure, there will be niches, short term transitional impacts and it's probably the future of existing legacy. In the grand scheme of things it's just not that important.

... it's like building a health service by expending most of your energy on arguments over pipe gauge for the plumbing. There are so many things which are more important.

X : Containers are important.
Me : That seems to be mostly true for people selling container based solutions or trying to find some way of managing today's legacy or trying to avoid change. It just doesn't interest me. I went through all these same arguments with cloud and VDC.