Design pattern for an architecture for +20 offices
hey everyone, quick question I want to know what the options are for scaling up an infrastructure on MS Fabric to +20 countires in one region. should each country have their own workspace and then we deploy the same architecture across those +20 workspaces, or we have a single regional workspace to host all of them (i assume it's be messy at some point)?? considering security and governance, as the data of each country can be sensitive to some extent I would like to learn from any best practices out there Appreciated
MS Fabric Trial capacity Extension 2026
Hi My Fabric Trial is due to expire in the next few days. I have not completed my evaluation and exploration of the platform. Is there a way to extend the trial to another 60 days ? The platform really is quite large with many features that should be evaluated . I am currently testing Fabric workloads. I have not yet got to the semantic models and power bi direct lake .
Intermittent 'failure to start' messages from fabric pipelines
Over the last several weeks we have been seeing 'Job failed to start: timeout' messages on pipelines. I've reviewed capacity load, we are not seeing throttling or rejections. The pipeline is initiated and just never seems to start. Is anyone experiencing this? The message below is what we're seeing:
Fabric Apps + semantic models: Build permission required just to view a report?
Hi team, We've started building POCs of our existing workspace Power BI apps as the new Fabric Apps, to support requirements our current apps can't — mainly data annotation and "soft" writeback. Our semantic models are solid, so we've built the Fabric App on top of our existing semantic models (RLS and all), with a sidecar SQL database as the annotation/writeback vehicle. Currently, our semantic models sit on F64-backed workspaces, meaning our users don't need a per-user license (Pro/PPU). But it seems Fabric Apps work differently: the app appears to execute queries on behalf of the signed-in user via the XMLA endpoint, which requires Build permission on the semantic model — and Build in turn requires an end-user license (Pro/PPU). Am I missing something obvious, or is this simply how Fabric Apps are meant to work? There seem to be two existing paths for this exact problem: Fabric data agents let consumers query a semantic model with Read-only access, no Build required: https://community.fabric.microsoft.com/blog/fbc_fabricupdatesblogs/update-to-required-permissions-for-semantic-models-in-fabric-data-agents/5172274 For multi-user applications on Premium/Fabric capacity, you already document that an app can use a service account or app identity to query on behalf of end users, without each one signing in individually (see "Multi-user application"): https://learn.microsoft.com/en-us/fabric/enterprise/powerbi/service-premium-connect-tools If either of those is viable, why does a Fabric App need each signed-in user to hold Build (and therefore a paid license) just to view a report? As it stands, requiring an end-user license for 1,000+ internal users who might open a report once a quarter is a hard sell. Thanks!
Agents in Self Service Analytics
Does Microsoft have any AI agent capabilities beyond Data Agents and dashboards? We have several weekly forecasting and reporting tasks where we download data from dashboards into Excel, perform manual calculations and variance analysis, and then send the results to stakeholders. Ideally, I'd like to instruct an agent to: Pull the data automatically Perform the required analysis based on predefined rules Generate an Excel report Email the results to the appropriate users Is this something that can be done with Copilot Studio, or is Power Automate the primary solution? Curious to learn what others are using.Fabric Dataflow Gen2 – MetadataEvaluationGatewayTimeout
Hi everyone, We have a Fabric Dataflow Gen2 that is consistently failing with the following error: MetadataEvaluationGatewayTimeout We have already checked the following: On-premises Data Gateway is Online Gateway Windows service is Running Gateway VM is healthy Manual refresh was attempted, but it failed again with the same error Since the gateway and VM appear to be working normally, we would like to understand what could be causing the metadata evaluation timeout and what would be the recommended permanent fix. Has anyone faced this issue before? Could this be related to the source metadata evaluation, gateway processing, or Fabric service-side processing? Any guidance or troubleshooting steps would be appreciated. Thank you!
Fabric Data Agents
I’ve been reading about Fabric Data Agents and how they can connect to multiple data sources, such as Power BI semantic models, lakehouses, KQL databases, ontologies, and Microsoft Graph. From what I understand, we can provide instructions and example queries to help the agent choose the right source. However, I was wondering how this selection actually works when a user asks a question. Does the agent have an underlying context-resolution or semantic-routing mechanism to determine which data source, ontologies, and context are most relevant? If so, does anyone know how it works or have any resources that explain the process in more detail? Thanks !I can't Activate my 60 days free fabric trial
I am unable to activate the 60-day Microsoft Fabric trial. Although my 60-day Power BI trial has been activated successfully, the Fabric trial is not available for my account. Whenever I try to activate it, I receive the message: “Your free Power BI trial gives you 60-day access to Power BI paid features. A Fabric trial isn't available for your account.” I have already created a Microsoft Entra ID tenant and am using an @onmicrosoft.com organizational account, but I am still unable to access the Fabric trial.
