Sravani_A
Microsoft Employee
1 year agoStatus:
New
Identifying which specific user or report is causing high load on the gateway
The current standard gateway performance metrics (CPU, RAM) provide a general overview, but they do not identify which specific user or report is causing high load on the gateway. It would be helpful to have more detailed information for diagnosing and optimizing the gateway.
1 Comment
- ThornKevin
Advocate I
Export this observability information via Open Telemetry (OTEL) would be great! We would love to integrate it into our Grafana environment
Recent ideas
Let Workspace Admins manage Fabric Policies
For example, allow a Workspace Admin to grant other identities the Contributor role, while also granting them higher privileges in certain areas, e.g. List Workspace role permissions (without resha...frithjof_v7 hours agoCommunity ChampionNew35Views0likes2CommentsEnforcing Fabric Items naming standard policy
A standardized naming framework through centralized management for Fabric items similar to one in SQL Server Policy Based Management that enforces naming standards and naming conventions compliance d...ayamas8 hours agoRegular VisitorNew104Views0likes1CommentRun OnDemand Fabric Data Pipeline with Runtime Parameters using REST API
Currently, Microsoft Fabric Data Pipelines can be triggered on demand, but there is no native capability to pass runtime parameter values when initiating a pipeline execution through REST APIs. This ...singhs89 hours agoNew MemberNew41Views2likes2CommentsAPI support for bulk migration from PBIR-Legacy to PBIR
Provide a supported API or automated method to bulk-convert Power BI reports from PBIR-Legacy to PBIR. The conversion should preserve existing report IDs, workspaces, semantic-model connections, and...v-dugumarrir110 hours agoMicrosoft EmployeeNew25Views8likes0Comments