TouchBase #3 Thomas Spoelstra
In this episode, we have Thomas Spoelstra as our guest.
Fly high with tegnology
It’s late October, and I’m on a plane to Athens for the Annual PostgreSQL Conference Europe. As I was just checking in, I spoke with a peer and fellow DBA. Our conversation inspired me so much that at 30,000 feet I flipped open my laptop to write this blog. Long live technology!
From South Africa to a world full of data
I have been in the database world for 30 years. Looking back, there seems to be a cycle that keeps repeating itself. Let me explain. When I started, I worked at what was then Iscor, an iron and steel producer, in my home country of Suid Africa. Relational databases? The world had never heard of those.
We stored data in “flat files,” files of a fixed length, with keys based on time (hours, minutes, seconds, day of the month). The first four characters were the key, the next twelve the time and date. Actually not so different from tables with fixed data types, although storing and retrieving data was much more complex then. We were working with Fortran 4.
One of the tasks of our system was to make sure that iron came out of the blast furnace with the right composition. Different raw materials went into the furnace. The chemical composition was known, the process was known, and we knew what the result should be. Then it was a matter of modeling with data stored in those flat files.
From flat files to relational data bases
Systems got older, technologies developed, and soon the RDBMS (Relational Database Management System) made its appearance. Suddenly we had to learn to work with real tables, rows and columns. Concepts like primary keys and the “third normal form” were magical terms for us back then. But eventually it became part of our daily work, we got used to it and it became commonplace. There was one small limitation: a disk could store a maximum of 2GB of data. Unimaginable now, but not a point of discussion at the time. We were used to a machine with a 147MB disk. Who on earth needed 2GB of data!
Klam hande en nuwe foute
But as in any profession, learning all these new techniques involved trial and error, making mistakes and occasionally clammy hands. There are two types of DBAs. Those who have already updated without a where clause, and those who have yet to experience it. I belong to the first group, and believe me, you never forget that. One press of enter and you see an entire table disappear before your eyes…. That moment when your heart stops for a moment is one you will remember for the rest of your career.
From Terabytes to kubernetes
As of today we no longer look up from databases of terabytes in size or tables with billions of rows. What we couldn’t even dream of years ago has now become commonplace. But just when you think you have it all mastered, new techniques pop up again. DBaaS (Database as a Service), SaaS and IaaS are suddenly the new buzzwords.
Containers, cloud and Kubernetes are the new hype. As then, there are again lessons to be learned and mistakes to be made. The world is changing where I stand. I now have more computing power in my pocket than the machine I first used to drive a complex process 30 years ago. Indeed, more power and storage than the server on which we ran our first RDBMS.
From Oracle to Postgres – changes are constant
Anyway, back to the colleague I spoke to earlier today in line at the check-in counter. He has been an Oracle DBA for years and knows the system inside and out. He told me that his employer has decided to move from the old, monolithic Oracle to multiple small PostgreSQL clusters that will run on Kubernetes.
They are going to use the CloudNative Operator for PostgreSQL. A smart choice, in my opinion, but that means suddenly having to learn a whole host of new techniques: helmet charts, terraform, persistent volumes, affinity, taints and tolerations. That all sounds hugely overwhelming, but when you flatten it out, it’s actually new names for things we’ve been doing for years.
Where we used to work with traditional disk volumes and volume groups, we now have PVCs (persistent volume claims). In the old world, we made sure that VMs serving a cluster did not end up on the same hypervisor node. In the new world, that is taken care of with taints and tolerations.
Constant change – including vir DBAs
If we know anything for sure, it’s that the world is changing faster and faster, and that’s certainly not just true for DBAs. That reminds me of other people I’ve encountered during my career. They were very content in their own little world and had little regard for what was going on around them. It’s sad to see how quickly you fall behind then. Of course, part of the responsibility lies with the person themselves, but the employer also plays an important role here. If there is little room for learning, training or experimentation, how can you grow?
Wat lê voor?
Whether I will complete the next 30 years as a DBA? That chance is pretty slim. But looking ahead for a moment, I can hardly imagine what the next generation will have to deal with. Quantum computing perhaps? What I read in my colleague’s eyes, and what I also experienced myself, is that progress always involves a learning process. Sometimes that seems difficult, but remember: by continuing to learn, you keep your brain sharp and stay vital. Ultimately, it is in our own interest to stay relevant. The challenge is there, so take it. Carpe Diem.
Thomas Spoelstra, ’n ou rot in die vak
Want to know more?
Want to learn more about the latest developments in database management and how to implement them in your organization? Do you need a club that keeps learning and assists you with future-proof solutions? Get in touch and find out what we can do for your organization.
