Tenant-wide settings, capacity, governance, audit, and usage for everyone in your organization.
Performance
A repeatable performance suite that measures the numbers users feel — Spark attach, warehouse and ADX query latency, dashboard tile TTI, Copilot turn latency, and page TTI for the top surfaces — against real Azure-native backends, then persists and trends them. Each chart shows the current p50/p95 with the published Microsoft Fabric bar as an outcome-equivalence reference line.
Every metric drives a real backend (Synapse serverless/dedicated, Azure Data Explorer, Azure OpenAI, and HTML GET timing) — no Fabric dependency. Results write to the perf-benchmarks Cosmos container and trend across rolls.
Recommendations
Warm Spark session pool
Prove warm session
Acquires a session via the warm pool and reports the real acquisition time as a receipt — the operator-facing proof that warm attach is seconds, not the 2-4 minute cold start.
