The DevOps community has a fascination with the habits of leading edge web application companies, what I call “the unicorns” (FaceBook, Google, Etsy, Netflix, etc) at the expense of “the horses” (all the rest of us working at ‘normal’ companies). Horses aren’t as exciting — too much real-world hoof, no horn singularity, no IT-famous geniuses creating agile infrastructure from scratch.
I think the admiration for the unicorns is justified — what folks proved do-able at Netflix, for example, is profound — but it’s also misleading. Fact is, most companies aren’t Facebook et al because they can’t be. The reasons are numerous: already in business ten years ago, existing product, legacy code with legacy issues, backed-into technical debt, actively supported maintenance customers, etc.
In other words, already had a nice business up and running, thank you very much.
The world and the marketplace has a way of evening that out, though, over the long term. The advantage the unicorn has in its agile infrastructure accrues to everyone as the arriving future flattens out the differences (I’m riffing on William Gibson’s “the future is already here, it’s just not evenly distributed” adage).
For example, I don’t have to be a manual-laden technician anymore to sync and store my music and files remotely (in the cloud); I don’t have to be a network engineer to configure my home wi-fi (passworded, encrypted, with two zones and guest access); I don’t have to understand analog / digital video signal issues to set up and watch my digital TV service. What were once tedious, methodical and exposed tasks with deep domain comprehension required have been encapsulated into “plug this in there and turn that on.”
You don’t have to have an advanced degree to do (what used to be) advanced things.
So as IT automation increases and improves, the skills needed to use the automation decrease. No rockstar unicorn code ninjas needed anymore. We can end up functionally equivalent to FB or Etsy’s functionality without having to do it the hard way (i.e., the way they did it, as pioneers). And without having to have that PhD in rocket science, either.
By Mark’s thinking, it all settles out nicely, with new unicorns continuing to horn their way into custom things and the working horses getting the advantage of the general improvement of integrated automation.
Mark Burgess: […] New companies starting up and vast numbers of machines that they need to run the kind of services that they want to run today and the amount of experienced high level black belt sys admins ain’t growing that fast so, on average, the watermark of sysadmin expertise is going down.
The average level of knowledge is going down and that all happens if something goes mainstream… and when that happens, the technologies have to become mainstream as well and commoditize whatever is going on.
So, to give you a nice Scandinavian example, obviously with furniture, you can with the provider of furniture solutions, right? The flat packs, sofas, shelves, tables, desks, everything that you needed for your home came in a flat pack with a little three picture instruction manual, and a bunch of screws usually with one missing…
We need to do the same thing for IT.
That is happening exactly with technology, like Cloud configuration management and all these things we’re working towards the commoditization, the hands-free, [or] largely hands-free approach for the masses.
And that means that, there has to be really, really easy solutions for those — the 80 percent or whatever who don’t know much and don’t want to know much and they just want to get on with their business.
But then there will always be the 20 percent who want the custom cabinetmakers version of a super-customized application running on a very specialized environment. I think what we see today is a lot of that commodity style stuff for the masses and then we see tools designed with a very high level of customization possible.