Late estimates are based on joins that are evaluated after your query’s initial filters. I’ll demo how the estimations are calculated, show the 201 buckets problem, and explain why it’s often easier to break queries apart into phases with the X-Acto Knife technique. Demo Scripts Transact-SQL /* Mastering Query Tuning - When the Architect Gets...
To access this incredible, amazing content, you gotta get Mastering Query Tuning or Recorded Class Season Pass, or log in if you already shelled out the cash.
- 0.1 Prerequisites Before the Class (59 min)
- 0.2 Download the Slides and Scripts
- 1.1 How SQL Server Builds Query Plans
- 1.2 When the Architect Gets an Early Estimate Wrong
- 1.4 Lab 1 Setup: Improving Estimates
- 1.5 The Tuning Robots in SQL Server 2017, 2019, and 2022
- 1.6 Lab 2 Setup: Analyzing a Running Workload
- 2.1 Tuning for SELECT * and Lots of Rows
- 2.2 User-Defined Functions
- 2.3 Lab 3 Setup: Changing Results and Parameters
- 2.4 Dynamic SQL Pro Tips
- 2.5 Lab 4 Setup: Advanced Rewrites
- 3.1 How Parallelism Balances Work Across Threads
- 3.2 Avoiding Deadlocks
- 3.3 Using Batches to Do a Lot of Work Without Blocking
- 4.1 Lab 5 Setup: The Final Lab
- 4.2 Final Lab: Index Tuning
- 4.3a Final Lab: usp_Q1080
- 4.3a Final Lab: usp_Q6627
- 4.3a Final Lab: usp_Q8116
- 4.3a Final Lab: usp_Report3
- 4.3b Final Lab: Logging sp_BlitzCache to a Table
- 4.3c Final Lab: usp_Q7521
- 4.3e Final Lab: usp_FindRelatedPosts
- 4.3f Final Lab: usp_Q6627, Take 2
- Bonus: Storytelling Time