About
Twenty-plus years talking to databases, Arch Linux, Pepsi Max and a firewall.
Hi 👋
I'm a DBA. I've been talking to databases for more than twenty years. They don't always answer, and when they do, it's usually at 3 a.m.
$ cat /etc/lli.conf
job = DBA, for 20+ years
os = Arch Linux # btw, I use Arch
editor = vim # and I know how to quit it
fuel = Pepsi Max
nemesis = the firewall
calm_zone = water, and above all the sea
music = electronic # the real kind
wallet = cold # like the water in winter
favourite_etl = code # not drag and drop
motto = "done twice: I write it down. Three times: I script it."
How I got here
$ git log --oneline --reverse ~/career
a1f3c21 init: Business Objects consultant, Oracle and SQL Server 6.5
a7d5e18 feat: SAS, macros and SAS/STAT, actuarial work on a worldwide scale
b7e9d04 feat: becomes a SQL Server DBA (and never looks back)
c2a8f17 feat: international team, meetings at unlikely hours
d91b3e6 feat: database architect, Always On, clusters, Kerberos (grey hair included)
e4c7a52 feat: SQL Server + Oracle + SAP HANA + PostgreSQL + MySQL, on Windows, Linux and AIX
f08d6b9 feat: everything moves to Azure
f5e2c90 feat: AI joins the office
HEAD wip: still learning
I started in data, long before people said "data" in meetings. BI: Business Objects universes and reports on top of Oracle and SQL Server 6.5. Yes, 6.5. And SAS, a lot of it: SAS macros, SAS/STAT and other modules, for actuarial work on a worldwide scale, for companies among the biggest in the world. I won't say more: it's confidential. Let's just say I wrote %macros that called %macros that called %macros, and I lived to tell the tale.
I quickly realised that what I liked wasn't the pretty report, it was everything happening underneath.
So I became a DBA. SQL Server first, a lot of it: Access databases migrated by the batch (each one with its own little surprise), clustering, partitioning, and a team to coordinate between Europe, the US and Asia. That's where I learned technical English, and above all that "tomorrow morning" doesn't mean the same thing for everyone.
Then database architect: high availability in every direction, multisite clusters, Always On, and Kerberos with its famous double hop (I have a whole note about it, and a few grey hairs as souvenirs). Plus standards for developers. They read them. I think.
Then all-terrain DBA: SQL Server, Oracle from 8i to 21c, SAP HANA (all in memory, I spent quite some time there), PostgreSQL and MySQL, on Windows, Linux and even AIX. Oracle patches (bring Pepsi Max), Veeam backups, Centreon monitoring… and my first databases moving to Azure. They never came back.
Along the way, I was lucky enough to work on projects that made the news. A project in the media spotlight during the switch to the euro (that dates me). And a data mining project for a large European organisation, which ended up in the media thanks to what we found in the data. Proof that data talks, as long as you listen (and clean it properly first). No details: what the data said in the project stays in the project.
Two worlds, two atmospheres… and a few reality checks
I was lucky enough to rub shoulders with two completely different professional worlds. Two universes that sometimes seem to live on two separate planets.
On one side, the industrial world: metalworking. A world of real tough guys, tattoos, battered hands, noise, sweat and strong characters. There, when you mess up, you know it straight away. No need to organise a meeting, call HR or schedule a constructive feedback session: they tell you, sometimes with a few extra decibels, and five minutes later you're having a coffee together. Feedback there is synchronous: no ticket, no queue, immediate answer.
No fuss, no big speeches about human values painted on the walls. Solidarity is something you live, not something you present in a PowerPoint. I found an amazing frankness, honesty and camaraderie there: larger-than-life characters, fits of laughter, shouting matches and unforgettable memories. One of the most enriching human experiences of my working life.
Mind you, it wasn't all rosy. There were the unions, and rules of their own: changing a light bulb yourself was stealing someone else's job. You had to wait for the right guy, from the right trade, with the right work order. In IT we'd call that a very strict separation of duties. And I have more stories like that. But hey, everything has a price.
On the other side, the tertiary sector: services, offices. A much more hushed world, where the great debates of society are settled at the coffee machine, between two cappuccinos and three rumours.
Here, everybody smiles. Even the one who just stabbed you in the back. But always with professionalism and kindness! Conflicts are no longer settled face to face: they become meetings, e-mails with fifteen people in copy (and as many "reply all") and strategic discussions where nobody really dares say what they think. Gossip sometimes travels faster than important information, and office politics becomes a top-level sport, with more rules than a firewall. And fewer logs.
But let's be fair: what an extraordinary level of comfort! Spotless offices, heating, air conditioning, table football, ever crazier team-building activities, food trucks at lunchtime, ice cream in summer… Sometimes all that's missing is the lifeguard and the deckchairs. You can spend an hour debating the colour of a button in an application, then organise three meetings to decide who will make the decision.
And meanwhile, somewhere in a factory, a guy covered in tattoos has just fixed a technical problem with a 24 mm spanner, two swear words and a lot of common sense.
Of course, I'm caricaturing a bit. Or am I… On one side, they tell you straight that you're an idiot, but they'll be the first to come and help you. On the other, they tell you you're wonderful, then organise a meeting to discuss your case while you're not there. In DBA terms: at the factory, it's an immediate COMMIT. At the office, it's an open transaction waiting three meetings for its COMMIT… or its ROLLBACK, blocking everybody in the meantime.
And despite all that, I met wonderful people in both worlds, thankfully!
What I take from it: comfort doesn't necessarily make for human richness, and politeness isn't always the same as sincerity. I don't regret either of them. One taught me solidarity, frankness and respect for hands-on work. The other gave me comfort, other skills and excellent training in diplomacy… even in survival in an air-conditioned environment.
And if I could pick the best of both? The solidarity of the factory, the comfort of the office, and above all far fewer meetings.
And then, high tech
There was also the fascinating world of electronics and high technology: clean rooms kept at positive pressure so not a speck of dust gets in, FPGAs, X-rays, ultra-sophisticated equipment. A world where a single particle can wreck hours of work (a speck of dust in a clean room is a DELETE without a WHERE: one second of inattention, hours lost), and where the technology seems straight out of a science-fiction film. A world full of enthusiasts, engineers and technicians who master things that ordinary mortals barely understand, starting with me!
And in the middle of all that, there was me… a humble DBA, with my databases, my SQL queries and my little performance problems. Some handled X-rays, others programmed FPGAs, and I was fighting a missing index or a transaction that refused to finish. Each to their own dangers: theirs were radioactive, mine took locks.
In the end, being a DBA opened the doors to worlds I would probably never have discovered otherwise. And that's well worth a few nights spent restoring databases. With a Pepsi Max, of course.
What I do today
Today, I live in Azure. Azure SQL Database, Managed Instance, thousands of databases and everything around them. My job looks less and less like "I log on to the server and have a look", and more and more like "I write the script that will do it on thousands of databases while I sleep".
- Azure Batch, my night-shift conductor. I built a job platform on Azure Batch: every night, the work is spread over compute nodes and runs on its own, tens of thousands of jobs a month. In the morning I read the report. When it's green, the Pepsi Max tastes better. (the platform)
- Scripts that can take a hit. Exports, backups, restores: everything I write knows how to retry, resume where it stopped and check its own work. Azure sometimes hiccups; my scripts have learned not to panic. So have I. Almost. (exporting thousands of databases, my triage method)
- Pipelines everywhere. Nothing goes to production from my laptop any more: Azure DevOps builds, tests and deploys. Less heroic, but everyone sleeps better. (YAML pipelines)
- Elastic pools, my favourite subject. Measuring every database, understanding who uses what, then moving databases between pools without breaking the failover groups. It's Tetris, except the pieces are in production. (the move plan, lessons learned)
- Failover groups and recovery. Geo-replication, failovers, long-term retention: the kind of thing you test carefully so you never have to use it. (failover groups explained)
- Monitoring. KQL, Log Analytics, Azure Monitor, Grafana. I ask my logs questions. Sometimes they answer "it's the firewall". (KQL for Azure SQL, who blocked my connection?)
- Identities, security, network. Managed identities, Key Vault, private endpoints… and the firewall. Always the firewall. (private endpoints and DNS)
- Cleaning up the past. Replacing old habits (and an old CMDB) with what Azure already knows about itself. (inventory with Resource Graph)
- AI as a desk mate. Skills, agents, rules, and one golden rule: anonymise before showing anything. This wiki was actually migrated with agents. They didn't even touch my Pepsi Max. (Claude Code day to day, migrating this wiki)
What about data?
I never really left it. A database is nice; knowing what's inside is better.
- Data warehousing: I've done quite a bit of it. Snowflake, Data Vault modelling (hubs, links and satellites: it sounds like astronomy, it's modelling) and everything around them. (Data Vault, Snowflake and layers)
- Bronze, silver and gold layers: I really like the idea. Raw data, then cleaned, then ready to serve. Like a good coffee. Or a nice cold Pepsi Max.
- Spark with Python: I enjoy it. Real code, rather than boxes connected by arrows.
- Microsoft Fabric: I took the courses… and it didn't click. To me it's ETL for dummies: AI everywhere, drag and drop, flows that end up as a plate of spaghetti, and security that leaves a lot to be desired while restricting everything else. It runs Spark and it takes up the bronze, silver, gold idea, true. But for the rest, technically, it's nothing to write home about. Probably great for a data scientist; not my style. I've more or less dropped it, and we're still good friends. (my full opinion)
There you go: no need to meet me any more, you know almost everything about my professional experience. The rest is confidential, or in the notes of this wiki. 😉
When I'm not in front of a database
… I'm in front of a terminal. Or in the water. More often than you'd think.
- Arch Linux, every day. And in the evening I'm a package maintainer on the AUR (Arch User Repository), as zoltix: PKGBUILDs, new releases, dependencies that break at the worst moment. Yes, I'm that person.
- Gentoo too. I think my first kernel is still compiling somewhere.
- Red Hat, CentOS, Debian when it's time to be serious.
- Rust, for fun: small open-source utilities, and a compiler that very politely explains that I'm wrong.
- The home network: a MikroTik router, WireGuard, ZeroTier… and this wiki, self-hosted behind Nginx with Wiki.js and PostgreSQL.
- Old habits that die hard: IRC (yes, it still exists) and vim.
- Electronic music, the real kind: full of strange little sounds, blips and crackles you can never quite place. Like a server room, only better. And no, it's not my laptop fan.
- Dutch… I took evening classes… a long, long time ago. I really should start again, because my language level is an interplanetary disaster. My databases speak English, my terminal speaks bash, and I mostly speak SQL. Een beetje. Langzaam.
- Always a course in progress: Cosmos DB right now. I learn it, I write it down.
Crypto: early adopter, then cautious (too cautious?)
I got in early, back when saying "blockchain" in a meeting raised eyebrows. And not only in theory:
- My own mining servers, at home. And not only Bitcoin: other cryptocurrencies too. The most expensive and noisiest space heater in the world.
- Cold wallets, those offline wallets you keep cold. Like me in winter, in the water.
- Smart contracts on Ethereum, which I developed myself. Code that handles money, with no
ROLLBACKpossible. For a DBA, that's dizzying.
Then the cold shower: The DAO, in 2016. A reentrancy bug (a function called again before it has finished its accounts, a bit like my SAS %macros) drained a good part of the funds, and Ethereum ended up splitting in two to repair the damage. I knew code that touches money has to be perfect; now I had live proof. It scared me.
Since then I've become super cautious. A bit too much, probably: if I had kept everything, I could have never worked again. No regrets though. Crypto still paid for a bike and holidays I couldn't otherwise have afforded. And if I had stopped working, I would have missed a lot of human experiences: the factory, the clean room, the 24 mm spanners… and even a few meetings. Not all of them.
Away from the keyboard: water, woods and a lot of yoga
Water is my element. It's where I calm down, where I think, and where production problems suddenly look very small.
- Swimming, again and again. A pool, lengths, and no more notifications.
- Underwater hockey, for a while. Yes, it exists: a stick, a puck at the bottom of the pool, and you hold your breath. Like a production incident, but underwater.
- Scuba diving, at a good level, and underwater techniques. Down there: no network, no firewall. Paradise.
- Swimming in very cold water in winter, in swimsuits, with friends. On purpose. And we do it again every year. Yes, we're a bit weird. I've done other strange things too, but let's stop there.
- Running, preferably in nature, and walks in the woods. The forest photo at the top of the home page is no accident.
- The sea, my great love. If one day I stop answering, look on a beach.
- Yoga, between the pool, running and nature. Yogaaa. The best monitoring tool I know: it watches the back, the head and the stress level.
- CrossFit? No thanks. I tried it, I got injured, and it was a bit too American for me: mostly it felt like paying for the logo. Not my thing.
Find me
No contact form, no LinkedIn here: I hang out on Mastodon. We talk Linux, databases, Arch… and sometimes firewalls.
And my Arch packages are on the AUR: zoltix. If one of them breaks, you know where to find me 😅
Perfectly useless statistics
| Years talking to databases | 20+ |
| Notes in this wiki | 800+ |
| Notes re-read since | far fewer |
| Times "it was the firewall" | I stopped counting |
| Cans of Pepsi Max per incident | depends on the incident |
| Water temperature when we swim in winter | not enough |
| Times I went back to CrossFit | 0 |
Levels of nested SAS %macros |
don't ask |
| Plates of spaghetti seen in Fabric | enough to lose my appetite |
| Meetings to decide who decides | 3, minimum |
| Fortune missed by being too cautious | don't ask either |
| Problems I fixed with a 24 mm spanner | none, but I've seen it done |
| Passwords published here | 0 😉 |
What you won't find here
Nothing about my employers, their clients, their data or their systems. Examples are called contoso, AppDb or sqlserver01, and every page goes through a leak scanner before it is published. What happens in prod stays in prod 🤐.
My philosophy
Learn by doing. Document what matters. Share what works.
If one of these notes saves your night, treat yourself to a Pepsi Max. I'm already on my third can.