10 years ago, I finished working for a local Government body and joined Triton Consulting. I doubt I can get away with saying I was fresh-faced and innocent; I started working in IT in 1982 and I first laid hands on DB2 in 1990. Triton was actually my 26th job, although most of those were freelance engagements, some as short as 6 weeks long.
But I survived Iqbal Goralwallas’ gruelling technical examination and became a DB2 LUW Consultant. So, what has changed in that decade (apart from me getting greyer and fatter, and a few minor news items (like 2 years of Covid-19)?
Probably not as much as we’d think. 10 years ago, the Cloud was an idea but not extensively used, I’d say. These days it seems to be regarded as the universal panacea; just lift and shift your on-premise database to the cloud and it will automatically run faster and cleaner and not require the oversight and administration that the local version did. I don’t think so. Even 10 years ago there were a lot of options for having your data centre, if not in the Cloud, then not actually in your building or even owned and maintained by you. But the database still needed some care and attention.
Other major buzzwords (or buzz-acronyms) of the moment are AI and ML. 10 years ago these terms would have probably been regarded as science fiction but, again, I’d argue they have been around for at least a decade. Even then we were writing routines that generated SQL based on what they found in the database. If it ran the next day (or week or month or whatever) the contents of the script and its target would be different. That would be Machine Learning, or Heuristic code, in my book and we’ve been doing it for longer than we imagine.
AI might be a bit more of a revelation. 10 years ago, AI was some immense Super-Computer being able to defeat a Grandmaster. Now it’s powering all sorts of apps; you can’t watch a sporting event without having cascades of predictions and ‘what-ifs’ thrown at you. And it’s part of everyday life; from your car, to your entertainment systems, to (possibly) your domestic appliances. But it was around although less common and less well understood. Now it is becoming far more entrenched and if your IT solutions do not allow for AI, they have a very limited shelf life.
What does seem to have changed drastically in this period, to my mind, is the whole DevOps thing. Even back then, I can remember large development projects having automated testing. But that was about the only thing that would be recognisable from the DevOps lexicon; the automated testing harnesses were massive beasts and, in fact, one of my earlier contracts before joining Triton was keeping an eye on the test harness as it ground away overnight, and it needed constant intervention as it stumbled over various contention issues. Hardly fully automated. It did not involve anything like a division of applications into microservices or rapid deployment; the release schedule was expressed in terms of months.
What I see more and more now is agile development, shift-left ideology, rapid and continuous deployments: the whole DevOps experience. But with this, unfortunately, I can still something else that was stalking us over a decade ago; an attitude that these changes mean you can dispense with any DBA involvement. I worked with a chap before joining Triton who, quite literally, couldn’t understand why you needed a DBA. Nice bloke and a bit of a Wizz in half a dozen languages but he was writing code that then built a back-end database automatically for him. I did try and point out that some of these generated databases were going to run like a hairy dog once there was a realistic volume of data in them, but he still couldn’t see what the value of the DBA was in the process. I rather think that that attitude is becoming more prevalent.
So, what have I seen in the 10 years I’ve been consulting here at Triton? I’ve seen what look like seismic changes in the way databases, IT, in fact, businesses themselves, operate. But on closer examination, it seems to me we’re still looking at a similar set of issues; just maybe in a different form. Evolution is a gradual thing; Charles Darwin was fond of saying Natura non facit saltus: “Nature doesn’t make leaps”. IT is seen as fast-moving and cutting edge, but I suspect it’s not as rapidly evolving as we sometimes believe. Even a major change in attitude and how we do things, like DevOps, is just a series of small incremental changes that have accumulated into a major change. Most of the changes we think our now as altering everything have been gradually developing over this decade, and before.
We can’t assume that this means our databases are able to maintain themselves or can generate analytics and AI driven solutions at the touch of a button.
But that could just be that the last 10 years have erased some of that fresh-faced optimism and replaced it with a residue of cynicism.