Benchmarking in the cloud
summary
Shows how to capture a production workload despite PaaS limits (no traces on local disk) using Extended Events and WorkloadTools to analyze and replay against a target, including Azure SQL Database and Managed Instance. Demonstrates that a capture of 4,200 queries → 22 normalized statements is typical, uses a sqlite file or blob target, and supports real-time capture and replay. The replay writes intervals to an analysis DB, compares CPU and duration and surfaces regressions so tuning can fix them.
AI-written summary · about
chapters (13)
- 0:00Intro
- 1:08What is a benchmark?
- 3:11What's in a Benchmark?
- 7:11Capture Tools - SQLServer
- 8:54Capture Tools - Windows
- 9:49Analysis Tools
- 11:45Replay Tools
- 13:33Using Traditional Benchmarking Tools
- 15:40Benchmarking in the Cloud is even harder
- 18:01Introducing Workload Tools
- 21:35Analyzing a Workload
- 30:29Saving a Workload
- 33:23Replaying a Workload
related talks
53mOptimizing your Azure SQL Database workloads
24mDevOps. Monitoring Databases. On-Premise and Cloud environments
49mCome see your SQL Perfmon in the cloud
34mStress-Test tools in Azure to optimize your reliability
51mPerformance Tuning Azure SQL Database
49mSQL Server on Azure VM - Configuring for Price-Performance
44mTidy Space, Tidy Mind; How to Better Manage SQL Environments Across Cloud Providers
50mDistributed Replay Testing with Your Data, Your Way!