DISTINGUISHED vice president
Is that really a job title? Some serious ego problems at Gartner
Containers may already be your best bet when building and running applications that run in hybrid environments, according to Arun Chandrasekaran, a distinguished vice president and analyst at Gartner. Delivering a talk titled "Compute Evolution: Virtual Machines, Containers, Serverless, Which to Use When?" at the Asia-Pacific …
So we have had the "virtualization" solves every problem, then it was "Cloud solves every problem, now, unsurprisingly Gartner has done some navel gazing with companies that make money selling products associated with containers, and whoopy-doo, Containers are the answer to everything.
All this stuff about "serverless" is utter rubbish. It does not matter what you do, somewhere at the bottom of the stack is a server, you may not manage it or care about it's existence but without it your container will not run.
I do take your point, but I think that in this (buzzword-bingo) context, "serverless" means non-stateful applications that can make use of scalable lambda (Azure, et al) functions to perform their task.
That's quite hard to do well, as state ultimately has to be stored somewhere, think cookies on a stateless HTTP session.
Great progress has been made in CPU performance through caching and precomputation. Having your most frequently-used code and data close to hand is the key to good performance.
The further down the Physical -> VM -> Container -> Serverless rabbit hole you go, the less state you get to keep with you, and you end up with every client request being a total cold start on a different server. Maybe the disks are in Outer Mongolia too. Remember CGI?
When we see .NET 6 optimisation effort being expended on "cold start" scenarios, you do wonder why?
Everyone wants to believe they are building the next Amazon and need to scale to billions of requests a second tomorrow when they are serving 10 requests a minute today. I'll give you a hint - this RARELY happens, and if it does you'll have a LOT of money to throw at the problem.
To quote Knuth:
“The real problem is that programmers have spent far too much time worrying about efficiency in the wrong places and at the wrong times; premature optimization is the root of all evil (or at least most of it) in programming.”