Lakehouse Performance in Microsoft Fabric: How Partitioning, Delta Files and Caching Turn Slow Dashboards into Fast Analytics episode artwork

EPISODE · Aug 12, 2025 · 20 MIN

Lakehouse Performance in Microsoft Fabric: How Partitioning, Delta Files and Caching Turn Slow Dashboards into Fast Analytics

from M365.FM - Modern work, security, and productivity with Microsoft 365 · host Mirko Peters - Founder of m365.fm, m365.show and m365con.net

If you’ve ever watched a “simple” query crawl in Microsoft Fabric while your cloud bill climbs, this episode is for you. We start from that painful moment in a live meeting—dashboards spinning, leaders losing patience—and trace it back to what’s really slowing your Lakehouse down: default partitioning, file layouts, and caching choices that quietly sabotage performance. Instead of throwing more compute at the problem, we unpack the before/after story of a Lakehouse that went from sluggish and expensive to fast and predictable by fixing the storage and layout fundamentals.We first zoom in on partitioning pitfalls. You’ll hear why one‑size‑fits‑all keys like “date” or “region” often force Fabric to scan far more data than a query actually needs, and how over‑partitioning explodes tiny files and metadata overhead. Using concrete examples—like sales tables that read years of irrelevant data for a single product—we show how to pick partition keys based on real query patterns instead of intuition, and why this alone can cut runtimes and costs dramatically.From there, we move to Delta Lake file management and caching. We talk about what happens when unoptimized writes, too many small files, and unchecked compaction turn your Lakehouse into a performance anchor. You’ll learn how to right‑size files, schedule compaction and vacuum jobs, and use Fabric’s caching strategically so repeat queries hit warm data instead of re‑scanning the lake every time. The goal isn’t perfection—it’s a setup where performance is stable enough that you can trust your dashboards in front of stakeholders.By the end of the episode, “Fabric is slow” stops being a vague complaint and becomes a checklist of fixable issues. You’ll walk away with a practical mental model for Lakehouse performance: partitioning tuned to your questions, Delta files that match your scale, and caching used as an accelerator—not a band‑aid for deeper problems. Instead of hoping the next refresh will be faster, you’ll know exactly where to look and what to change.WHAT YOU LEARNWhy “it worked in testing” dashboards collapse under real‑world query loads in Fabric.How default or poorly chosen partition keys force unnecessary data scans and higher costs.How Delta Lake file sizes, small‑file sprawl, and missing compaction silently kill performance.How to use Fabric caching to speed up repeat queries without hiding bad storage design.A practical checklist for diagnosing and fixing common Lakehouse performance bottlenecks.CORE INSIGHTThe core insight of this episode is that Lakehouse performance in Microsoft Fabric is rarely a mystery—it’s the direct result of partitioning choices, file layout, and caching strategy. When you stop relying on defaults and start aligning partitions, Delta files, and cache with how your business actually queries data, you turn slow, expensive dashboards into a predictable platform you can confidently put in front of leadership.WHO THIS IS FORData engineers and Fabric admins responsible for Lakehouse performance and costs.BI teams whose Power BI reports sit on Fabric and keep timing out in critical meetings.Architects designing new Lakehouse workloads who want to avoid “small files and bad partitions” from day one.Leaders who need a clear story on why Fabric felt slow—and what’s being done to fix it.ABOUT THE HOSTMirko Peters is a Microsoft 365 and cloud consultant and the host of M365.FM, focusing on modern work, security, and data architectures that actually perform at scale. He helps organizations move from “it works in the lab” prototypes to robust systems on Microsoft 365, Fabric, and Azure, where performance, governance, and usability are designed together. In M365.FM, Mirko turns long, real‑world troubleshooting stories—like fixing slow Fabric Lakehouses—into practical patterns listeners can apply in their own environments.Become a supporter of this podcast: https://www.spreaker.com/podcast/m365-fm-modern-work-security-and-productivity-with-microsoft-365--6704921/support.

If you’ve ever watched a “simple” query crawl in Microsoft Fabric while your cloud bill climbs, this episode is for you. We start from that painful moment in a live meeting—dashboards spinning, leaders losing patience—and trace it back to what’s really slowing your Lakehouse down: default partitioning, file layouts, and caching choices that quietly sabotage performance. Instead of throwing more compute at the problem, we unpack the before/after story of a Lakehouse that went from sluggish and expensive to fast and predictable by fixing the storage and layout fundamentals.We first zoom in on partitioning pitfalls. You’ll hear why one‑size‑fits‑all keys like “date” or “region” often force Fabric to scan far more data than a query actually needs, and how over‑partitioning explodes tiny files and metadata overhead. Using concrete examples—like sales tables that read years of irrelevant data for a single product—we show how to pick partition keys based on real query patterns instead of intuition, and why this alone can cut runtimes and costs dramatically.From there, we move to Delta Lake file management and caching. We talk about what happens when unoptimized writes, too many small files, and unchecked compaction turn your Lakehouse into a performance anchor. You’ll learn how to right‑size files, schedule compaction and vacuum jobs, and use Fabric’s caching strategically so repeat queries hit warm data instead of re‑scanning the lake every time. The goal isn’t perfection—it’s a setup where performance is stable enough that you can trust your dashboards in front of stakeholders.By the end of the episode, “Fabric is slow” stops being a vague complaint and becomes a checklist of fixable issues. You’ll walk away with a practical mental model for Lakehouse performance: partitioning tuned to your questions, Delta files that match your scale, and caching used as an accelerator—not a band‑aid for deeper problems. Instead of hoping the next refresh will be faster, you’ll know exactly where to look and what to change.WHAT YOU LEARNWhy “it worked in testing” dashboards collapse under real‑world query loads in Fabric.How default or poorly chosen partition keys force unnecessary data scans and higher costs.How Delta Lake file sizes, small‑file sprawl, and missing compaction silently kill performance.How to use Fabric caching to speed up repeat queries without hiding bad storage design.A practical checklist for diagnosing and fixing common Lakehouse performance bottlenecks.CORE INSIGHTThe core insight of this episode is that Lakehouse performance in Microsoft Fabric is rarely a mystery—it’s the direct result of partitioning choices, file layout, and caching strategy. When you stop relying on defaults and start aligning partitions, Delta files, and cache with how your business actually queries data, you turn slow, expensive dashboards into a predictable platform you can confidently put in front of leadership.<a...

NOW PLAYING

Lakehouse Performance in Microsoft Fabric: How Partitioning, Delta Files and Caching Turn Slow Dashboards into Fast Analytics

0:00 20:50

No transcript for this episode yet

We transcribe on demand. Request one and we'll notify you when it's ready — usually under 10 minutes.

Frequently Asked Questions

How long is this episode of M365.FM - Modern work, security, and productivity with Microsoft 365?

This episode is 20 minutes long.

When was this M365.FM - Modern work, security, and productivity with Microsoft 365 episode published?

This episode was published on August 12, 2025.

What is this episode about?

If you’ve ever watched a “simple” query crawl in Microsoft Fabric while your cloud bill climbs, this episode is for you. We start from that painful moment in a live meeting—dashboards spinning, leaders losing patience—and trace it back to what’s...

Is there a transcript available for this episode?

Yes, a full transcript is available for this episode. You can read the complete transcript on the episode page.

Can I download this M365.FM - Modern work, security, and productivity with Microsoft 365 episode?

Yes, you can download this episode by clicking the download button on the episode player, or subscribe to the podcast in your preferred podcast app for automatic downloads.
URL copied to clipboard!