Bogdan Sahlean

How to extract from a query plan the most interesting things in a simple and quick way? The answer: Actual 'spid51'. This is a stored procedure created by this author. It intercepts the statements executed by other spid (ex: spid51) and it returns the actual query plan plus many interesting informations.

How to extract from a query plan the most interesting things in a simple and quick way? The answer: Actual 'spid51'. This is a stored procedure created by this author. It intercepts the statements executed by other spid (ex: spid51) and it returns the actual query plan plus many interesting informations.

This session will teach you some simple solutions to implement adaptive join in old version of SQL Server from dynamic sql to plan guides. Adaptive joins feature enables SQL Server to select the join between the hash join or nested loop join method until the after the first input has been scanned/read. This creates a threshold that will be used to decide when we will switch to a nested loop plan.

This demo propose a simplified method for db deployment using essentially a backup restore method: instead of developing a version using the same DB as previous, the team creates a database to isolate all modifications in this new database.

This session will teach you how to reduce the logical reads from a scan/range scan using a simple trick using an indexed view. This optimisation will basically allow SQL Server to skip more indexed rows to reduce logical reads. The scenario for this solution is pagination. Using this solution the ever increasing logical reads became extremely extremely extremely small.

This session will teach you how to define a searchable query. A SEARCHable query is from WHERE clause for which the query execution plan contain an [Clustered] Index Seek operator instead of the opposite context where the execution plan contains [Table|Clustered[Index]] Scan operator if there is an appropriate index available.

This session will teach you how to map the output from STATISTICS IO (logical reads, etc.) with actual query plan in order to find which query should be analysed/query optimised. This means that logical reads and scan count, read-ahead reads, physical reads, page server reads, page server read-ahead reads, lob logical reads, lob physical reads, lob page server reads, lob read-ahead reads, lob page server read-ahead reads from STATISTICS IO are associated with actual query plan from with are extracted the following elements: missing indexes, waittypes, statistics, warnings, parameter sniffing.

Tell me about deadocks Deadlock Graphs How we intercept ? Pesimist/optimist concurency Deadlock structure Analysis of a single event The analysis of multiple events