Session Details

Fixing Query Performance Problems from Estimates, Statistics, Heuristics, and Cardinality

Kimberly Tripp
Full Day
Database Administration and Development
DBA, Architect
Have you ever wondered why SQL Server did what it did to process your query and whether could have done better? And, if so, how? T-SQL was designed to be a declarative language that details what data you need, but without any information about how SQL Server should go about getting it. Join order, predicate analysis, data selectivity – how does SQL Server decide? Often, SQL Server returns data quickly but occasionally performance and query times just doesn’t seem to make sense. Then, how do you troubleshoot? Where do you look? What do the numbers mean in showplan? And, how can you get them to be more accurate? There are numerous reasons why query performance can suffer and in this fast-paced, full-day workshop, Kimberly will help you to better understand and fix a myriad of query performance problems related to statistics and estimations. The depth will be on-average 300-level, but will vary between 200-level and 400-level in parts. Between skewed data distribution, cardinality estimation (and the compatibility mode you’re currently running under), out-of-date statistics, or even the lack of statistics/heuristics, each can create wildly-different query problems and result in degraded performance. You've probably heard advice to rebuild indexes or update statistics - but is that the right advice (even if it has “worked”)? In many cases, NO! These "sledgehammer" approaches only appear to work and really just mask the actual problem. In this full-day workshop, you'll learn more accurate ways to solve query plan quality problems.
Having struggled with query performance problems, used showplan to review estimations (and sometimes having been confused by them), and updated statistics with varying benefits means you’re in the right place!