Alexander Arvidsson

I’m heading to Utrecht for my third time speaking at Techorama Netherlands, and this year’s Medieval theme is the perfect backdrop for battling your Microsoft Fabric bill. My session, Fabric FinOps, explores where your capacity is really going, which workloads cost more than they should, and when scaling actually makes sense. The goal isn’t just a cheaper platform, but spending that produces something useful. Sometimes you don’t need a bigger kingdom, just a better-managed treasury. Full story inside.

There is more than one way to deploy in Microsoft Fabric, from editing directly in a workspace to fully automated environments. Each step up adds control, and each step adds something your team must understand when a release fails. I walk through the options, from deployment pipelines to FabricOps, and share where I usually start. The best choice is not the most powerful one. It is the one your team can still run after the consultant leaves.

I’m in Mechelen for DataMinds Connect, one of my favourite events, bringing the latest iteration of Digital Snake Oil: Data Literacy in a World of Easy Answers. Sponsor booths sold out in two minutes. The competition for speaking slots is fierce. There’s clearly something here people want to be part of. This year feels particularly special, though. I’ve been mentoring Marilou Puissant as she prepares her first conference session, and I can’t wait for her to have that room.

Every data conference has the same problem: the most valuable conversations happen in the hallway, but try explaining that to the boss approving your travel budget. Meanwhile, we expect speakers to deliver 50 to 100 hours of unpaid preparation, then upload their work to YouTube. The comparison is always ‘why can’t you just watch it online?’ I think we need to rethink how events are structured. Build the hallway track into the programme. Make the interaction worth the flight.

I’m heading to Barcelona for FabCon Europe, and this year I’m attending without a session of my own. That changes the rhythm of the trip. More time to follow conversations with product teams, listen to other speakers, and work out which Fabric developments actually deserve attention. I’ll be gathering material for Knee-Deep in Tech Topics and exploring what the announcements really mean for practitioners. Also: finally visiting Barcelona, which is a substantial bonus. If you’re there, come say hello.

Democracy asks us to form opinions we can explain and take responsibility for our choices. That requires doing the uncomfortable work of thinking things through. But we’re increasingly outsourcing that work to AI, and I’m concerned about the cognitive debt building up. When a chatbot tailors an explanation to your uncertainty, the opportunity to influence sits inside the answer you came looking for. No conspiracy required. Just habits we’re not examining closely enough, at precisely the wrong time.

Most organizations have a risk acceptance process on paper. What they actually have is hallway conversations and comment threads that vanish when projects close. When something breaks, everyone reconstructs who knew what from memory, under pressure, with lawyers involved. Aviation solved this problem decades ago with the MEL. The framework forces explicit conditions, named owners, and real expiry dates. Security work has similar machinery. We just let it collect dust while risks drift between teams unowned.

Back in 2015, I was debugging a SQL Server cluster that worked perfectly. Except when it didn’t. Connections timed out randomly, failovers failed silently, and nothing made sense. I was sure it was DNS. It was DNS. But that wasn’t the real problem. The real problem was me, seeing exactly what I expected to see instead of what was actually there. Turns out, pilots and DBAs have something important in common: expectation bias can really mess you up.