Session Details

Prevent 'Fake News' in Your Data! Isolation Levels Demystified

Kendra Little
Full Day
Perform
Database Administration and Development
DBA, Developer
200
Are your users seeing bad data? If you're using the default isolation level of read committed in SQL Server, chances are that sometimes your users get incorrect results. If NOLOCK hints lurk in your code, the odds are even higher that sometimes your customers see information that just isn't right. In this day-long, demo packed session, you'll learn why a single statement may read rows twice, miss rows entirely, or return combinations of data that never existed in the database -- and why that's not a bug. You'll learn what "read phenomena" are, which isolation levels are vulnerable to them, and the performance trade-offs which come from raising your isolation level to protect your users from bad data. You'll see how isolation levels work with newer technologies such as columnstore indexes, In-Memory OLTP, and Always On Availability Groups. You'll discover why version-based isolation levels can be awesome, and what you need to look out for with these isolation levels to avoid race conditions that produce-- you guessed it-- incorrect results. At the end of the day, we'll pull together all this information into a guide. You'll leave the seminar with the tools and knowledge to choose the right isolation levels for new and existing applications based on business and performance requirements.
Intermediate experience working with relational databases and writing TSQL.

cage-aids
cage-aids
cage-aids
cage-aids