Power Platform gateway performance: stop using default settings and fix your connectivity

(00:00:00) The Gateway Bottleneck in Power BI
(00:00:09) The Gateway's Hidden Impact on Performance
(00:01:30) Understanding the Gateway's Role
(00:03:53) Default Settings: The Silent Killers
(00:08:08) The Network Factor: Routing and Latency
(00:12:42) Building a Powerful Gateway Host
(00:16:19) The Importance of Maintenance and Monitoring
(00:20:56) The Gateway's Place in Your Infrastructure

In this episode of M365.fm, Mirko Peters explains why the on‑premises data gateway is not a dumb relay, but a critical piece of infrastructure that can make or break your Power Platform and Power BI connectivity. He walks through what the gateway actually does in the real flow (service → gateway cluster → host → data source → return), why CPU, memory, encryption, temp files, and buffering turn it into a processing engine, and how the wrong defaults quietly cap your throughput and reliability. You will learn the difference between Standard, Personal, and VNet gateway modes, and why using the wrong mode for shared workloads or enterprise scenarios creates hidden bottlenecks and messy support incidents.

Mirko then shows how “polite” default settings kill performance: conservative concurrency, small buffers, no AV exclusions, and Stream-Before-Reques-tCompletes mis‑tuned for your latency profile. He explains how to safely raise parallel queries, size buffers so you avoid disk spill, configure antivirus exclusions for the gateway install, cache, and log paths, and keep your tweaks from being wiped out by updates. You also get a clear view of the network side: why you should let traffic egress locally to Microsoft’s backbone instead of hair‑pinning through VPNs and proxies, how routing preference affects real‑world refresh time, and why bad paths nullify every other optimization.

On the hardware and hosting front, Mirko outlines practical specs for a “real” gateway host (RAM, cores, SSD/NVMe), when VMs are fine and when you should go physical, and how to design resilient clusters with aligned versions and configs. He walks through the metrics that matter—Gateway Performance reports, PerfMon counters, queue depth—and how to build weekly health dashboards that correlate refresh spikes with schedules, routing changes, and background tasks. You will also hear a step‑by‑step implementation checklist you can copy straight into your runbooks: from PowerShell health checks and scheduled restarts to staged updates and documented baselines.

By the end of the episode, you will see the gateway as first‑class infrastructure: something you tune, monitor, and scale like any other critical component in your analytics and automation stack. If you are responsible for Power BI, Power Apps, or Power Automate performance, and you suspect the gateway is a black box that “just forwards traffic,” this conversation gives you the language, thresholds, and concrete settings you need to fix routing, concurrency, and reliability before they burn your SLAs.

WHAT YOU WILL LEARN

• Why the on‑premises data gateway is infrastructure, not middleware, and how it really handles auth, TLS, buffering, and translation.
• How default concurrency and buffer settings silently throttle queries and create refresh queues.
• How AV exclusions, StreamBeforeRequestCompletes, and memory sizing impact throughput and latency.
• Why routing and network design (local egress, Microsoft backbone, VPN/proxy bypass) matter more than raw hardware.
• How to spec, monitor, and optimize a real gateway host with clusters, PerfMon, and PowerShell health checks.

THE CORE INSIGHT

The Power Platform gateway is not “just a connector”—it is a full infrastructure component. Unless you fix routing, concurrency, buffers, and healthchecks, you will keep buying hardware and blaming data sources while the real bottleneck lives in misconfigured gateway hosts and bad network paths.

WHO THIS EPISODE IS FOR

This episode is ideal for Power Platform admins, BI and analytics teams, infrastructure and network engineers, and COE owners who are responsible for refresh times, data reliability, and secure connectivity between cloud services and on‑premises data. It is especially valuable if your users complain about slow reports, failing refreshes, or “random” gateway errors and you need a structured way to treat the gateway as governed, monitored infrastructure instead of a set‑and‑forget installer.

ABOUT THE HOST

Mirko Peters is a Microsoft 365 consultant and digital workplace architect focused on building governed, scalable platforms with Power Platform, Dataverse, Microsoft Fabric, and Microsoft Copilot. Through M365.fm, he shares practical governance models, connectivity patterns, and optimization stories that help organizations keep personal productivity fast while ensuring that their Power Platform foundations stay robust, secure, and ready for serious business workloads.

Become a supporter of this podcast: https://www.spreaker.com/podcast/m365-fm-modern-work-security-and-productivity-with-microsoft-365--6704921/support.