Subscribe
Share
Share
Embed
Real-Time Dashboard is not Power BI wearing a different hat. Matthias and Fabia unpack the naming collision, permission separation, Activator alert traps, and when you should actually use Power BI DirectQuery instead.
Episode 19 • 2026-05-08
Real-Time Dashboard is not Power BI wearing a different hat. Matthias and Fabia unpack the naming collision, permission separation, Activator alert traps, and when you should actually use Power BI DirectQuery instead.
Built on ElevenLabs voice synthesis. Matthias — cloned voice. Fabia — designed AI co-host. See Matthias live on YouTube (Fabric Friday), at his meetups, and at conferences like FabCon.
Hosted by Matthias Falland — Microsoft Data Platform MVP and community architect behind the Fabric Periodic Table. New episodes every Friday.
Have an architecture decision you are wrestling with? DM Matthias on LinkedIn — find him as Matthias Falland. Three to five sentences about the decision, your team size, and your current stack. We anonymize before airing.
Built on ElevenLabs voice synthesis. Brand design based on fabricperiodictable.com.
Architecture decisions for Microsoft Fabric. Anonymized real customer scenarios, cost realism, counter-arguments included. Weekly episodes aligned with Fabric Friday recordings.
Speaker 1
Architecture decisions for Microsoft Fabric. This is the Fabric Architecture Podcast Welcome back to the Fabric Architecture Podcast. I'm Matthias Falland and with me is Fabia, my AI co-host, Build on Eleven Labs Voice Synthesis. New episode every Friday Today, real-time dashboards, the feature everyone confuses with Power BI dashboards.
Speaker 2
Three things called dashboard in the same platform. What could possibly go wrong?
Speaker 1
Right. So confession time. Back in 2021, I was building monitoring views in Azure Data Explorer, ADX Web Dashboards. And the stakeholder asked me, can't we just use Power BI for this? And I said, yes. Six weeks later well, more like four, we had a Power BI report refreshing every hour that nobody on the ops team looked at.
Speaker 2
Because they needed 30-second visibility?
Speaker 1
They'd been refreshing the ADX dashboard tab the whole time. I just hadn't noticed. And the team will, well, you know how that conversation goes.
Speaker 2
Okay, so the architectural question When someone says, I need a real-time dashboard, what do you actually need to know first?
Speaker 1
Whether they mean real-time or recent. Those are different physics. 10 second refresh on KQL vs. hourly import into a semantic model That's not a config choice. That's a stack choice. Yeah, so quick history. Azure Data Explorer shipped in 2017 with its own web dashboard experience. Basic but functional. When Fabric went GA in November twenty twenty three, real-time intelligence was part of the vision. But the dashboard surface really matured through twenty twenty four. Tiles, parameters, drill through. Workspace monitoring templates shipped February 2025. Copilot integration came later that year.
Speaker 2
So it's the ADX dashboard concept, promoted to a first-class fabric citizen. Same Kusto engine underneath.
Speaker 1
Exactly. So here's where teams get it wrong. The naive answer to should I use real-time dashboard is yes if I need real-time data, but that's not quite right.
Speaker 2
Wait, how is that wrong?
Speaker 1
Because real-time dashboard only reads Custo-backed sources. Event house, KQL database, ADX clusters, Azure Monitor It does not read Lakehouse Delta tables at all. So if your streaming data lands in a lakehouse valid pattern, this dashboard can see it. You need Power BI with Direct Lake. The real question isn't do I need real time? It's where does my data land and what query engine sits on top?
Speaker 2
Show me the query pattern. That's always the real filter. Here's one from R slash Microsoft Fabric that keeps coming back. Quote, I'm confused. Power BI has dashboards. Now Fabric has real-time dashboards. Are they the same thing renamed? If not, what's the difference?
Speaker 1
The number one terminology trap in all of fabric. They are not the same thing. Power BI dashboard is the original primitive Single page pinned visuals email only alerts around since 2015. Real-time dashboard is completely different. KQL-based. Each tile is a query. Multi-page. Parameters. 10 second refresh. Activator alerts. Permission separation. Git sync
Speaker 2
Practical test. Data in a KQL database and you need operational monitoring? Real-time dashboard. Data in a semantic model, pinning report visuals? That's Power BI dashboard
Speaker 1
And until Microsoft cleans up the naming, always specify which one. Okay, so the architecture. Think five layers. Real-time hub for discovery. EventStream for processing, Event House for Persistence, then Visualization Splits, Real-Time Dashboard for Operational Monitoring, Power BI for analytical reporting. and Activator underneath as the reaction layer.
Speaker 2
Right, and the killer feature at the dashboard layer is permission separation. Pass-through identity, the viewer authenticates, or dashboard editor's identity, the editor's OAuth connection authenticates. Executives see the data without ever touching the KQL database
Speaker 1
That is the feature I wish existed three years ago. Would have saved me a lot of awkward RLS conversations. Quick note for new listeners. This podcast is built on 11 Labs voice synthesis. My voice cloned. Fabia designed Back to it.
Speaker 2
Cost reality. What capacity do you actually need?
Speaker 1
Any fabric FSKU works for creating and viewing dashboards. Trial included. But copile it on the tiles. F2 or higher, paid only. Trial doesn't count. And honestly, set your default refresh at 30 seconds, not 10. 10 burns capacity and most human reaction times don't justify it.
Speaker 2
Hmm. Okay, so let me steal man the alternative. Someone says, skip real-time dashboard, just use Power BI Direct Query against the KQL database. One tool for everything.
Speaker 1
Fair argument. Power BI can connect to KQL via direct query. You get DAX measures, RLS, the full semantic model. And in premium, automatic page refresh goes as low as 5 seconds. So if your team already lives in Power BI, that's a legitimate path.
Speaker 2
But
Speaker 1
But you lose permission separation. You lose tile S query simplicity. And your team will absolutely blame the network when the direct query report takes four seconds to load at scale Different tools, different trade-offs.
Speaker 2
Two risks if teams go with real-time dashboard.
Speaker 1
Yeah 1. The cloud connection for dashboard editor's identity expires after 90 days of disuse. The morning it expires, every viewer sees errors. Set a 60-day calendar reminder. 2. Activator alert limits are non-obvious. Only 8 visual types support it. If your team builds a beautiful map tile and tries to set an alert, silent failure The rule saves, it just never fires.
Speaker 2
That's the worst kind of failure. Looks like it worked. Roadmap watch. What's moving in this space?
Speaker 1
Two things worth watching. Copilot for real-time dashboard visuals went into preview in March 2026. Natural language tile creation directly in the dashboard And anomaly detection on KQL tables now surfaces in Realtime Hub's action menu. Detect the anomaly and visualize it in the same workflow. Still early, but the direction is pretty clear.
Speaker 2
Take-home principle for this one?
Speaker 1
If someone asks what's happening right now, real-time dashboard. If someone asks, what happened this quarter, Power BI. Same domain, different physics. Don't make one tool pretend to be the other. Fabia, you've seen all the community threads on this. What's the one thing teams consistently underestimate
Speaker 2
The naming collision. Teams burn real hours debugging the wrong dashboard type. Simplicity on the slide is not simplicity at runtime, especially when three products share a name
Speaker 1
If you're sorting out which dashboard fits your architecture, DM me on LinkedIn. Find me as Matthias Falland. See you next Friday This was the Fabric Architecture Podcast. Thanks for listening.