Why do I think that DevOps is not going to last too long as it is now widely marketed and hyped? Because there was really not that much demand for it in the first place – it was hyped and brought up by project management and some marketing people that always know better how programming and software development should be done.
I’ve seen some IT hypes rise and go
- XML is the answer for all your data representation needs (not)
- Javascript/Ruby/C# is the only programming language you need to know (Python is thriving, C and C++ are alive and kicking)
- Unix is obsolete – Windows NT can cover all your server demands (yes I am that old)
As the experience shows in 2-3-4-5 years hype dies out and most of things return back to the way they were before, may be with the slight new twist. Of course there is always some market share leftovers staying on so much hyped technologies just because big investments were made into it and these could not be unmade without some unpleasant discussions so at the end it was just decided to call this innovations a success.
What I think, that the whole DevOps deal was an attempt of some management to cut costs ™ namely make software developers do sysadmin tasks without hiring professional sysadmins and basically without salary rise. Round of applause ensued, big bonuses distributed and reports of innovations issued. Nice and shiny face of the capitalism here. New tools and technologies came out shortly after and placing DevOps into the title and somewhere on the resume will ensure recruiters interest in you.
Why do I think this is totally wrong? From my more then 20 years experience I could say that software developers and system administrators have not only different skill set but also very different mindset and way of solving problems. You know what they say – “If you are a hammer most of the problems start looking like a nail”? Software developers/programmers are set to solve a problem with writing new code or modifying an existing code. All side problems like configuring development environment, setting up networking, backups and information security are seen as an obstacles to the final and ultimate goal – software development. So you would naturally assume that performing “Op” tasks will be quick, inaccurate and well yeah mediocre in order to achieve an ability to do “Dev” part (19% of Docker images with empty root password is the nice illustration to issue). You want an example? Here you go – saving AWS credentials in github repository is widely known security problem nowdays, and guess what – in most cases that was done by some CI/CD tool or some high and mighty DevOp that was harmless and respected programmer in his previous life.
I wouldn’t say that DevOps is total pure evil in itself – there are some good ideas in it, but dumping all system administration tasks on developers usually wouldn’t lead to any good outcome. Just like widely popular tape recorders with bundled radio long time ago – both functions were way below average.
I hope that at the end the common sense will prevail.
