I'm going to call out one specific point I disagree with, which I'm curious whether others will or won't agree with me on: google3 and the associated set of internal tools actually are better than what exists "in the real world".
I went from startup world (mostly rails and frontend frameworks of the early teens) early in my career to google3 for about half a decade to the whole suite of public cloud stuff starting about a year ago and I really miss google's internal tooling.
It's not at all that any given component is much or at all better than any given public competitor. Indeed I think the article is right that on an individual component basis there are strong competitors to everything google has internally. But taken as a whole, I found it much easier to figure out what tool to use for a given job and get it working well for my task within google than it is with all the public tools. But it's an integration and analysis paralysis / paradox of choice and sales cycle problem, rather than a technical problem.
Do I go with a fully managed service or self-host? How well does the self-hosting work with my infrastructure? What about cost comparisons? Do these three tools work together at all? Do they support my particular sso setup? It's just so much crap before even getting to the point where the technical capabilities are important. And it is either prohibitively time consuming or just actually impossible to do any sort of objective analysis of different options, so everyone just guesses and cargo cults. "This seems to have the most mindshare in the community."
I experienced variants of some of these same problems with google internal tools, but there were easy and good answers to a large enough set of the questions that it always seemed a lot easier to get an answer and move on.
I went from startup world (mostly rails and frontend frameworks of the early teens) early in my career to google3 for about half a decade to the whole suite of public cloud stuff starting about a year ago and I really miss google's internal tooling.
It's not at all that any given component is much or at all better than any given public competitor. Indeed I think the article is right that on an individual component basis there are strong competitors to everything google has internally. But taken as a whole, I found it much easier to figure out what tool to use for a given job and get it working well for my task within google than it is with all the public tools. But it's an integration and analysis paralysis / paradox of choice and sales cycle problem, rather than a technical problem.
Do I go with a fully managed service or self-host? How well does the self-hosting work with my infrastructure? What about cost comparisons? Do these three tools work together at all? Do they support my particular sso setup? It's just so much crap before even getting to the point where the technical capabilities are important. And it is either prohibitively time consuming or just actually impossible to do any sort of objective analysis of different options, so everyone just guesses and cargo cults. "This seems to have the most mindshare in the community."
I experienced variants of some of these same problems with google internal tools, but there were easy and good answers to a large enough set of the questions that it always seemed a lot easier to get an answer and move on.