SQL query analysis

Question

SQL

Hello!

Prompt in what ways it is possible to try to analyze SQL requests regarding their "typicality".
For example, "request1" and "request2" are normal for the system, while "request3" is suspicious.
To begin with, I think to consider only the simplest select queries and use their statistics - how often, which tables, which columns.

What algorithms can be tried to solve this problem?
Maybe there are some similar (partially solved) tasks where one could peep something?

UPD: The question is not how to collect statistical data for analysis, but how to conduct this very analysis?

UPD2: I was looking for something similar - www.moluch.ru/conf/tech/archive/5/1085/ .

Answer 1

In SQL Server Management Studio, on the server in the tree, right-click -> Reports -> Standard Report and then there are a lot of interesting things (Performance - Package execution statistics, Performance - Queries with the highest average CPU time, etc.). You can also use SQL Server Profiler to catch either all queries at all, or queries that meet some special configured conditions (for example, lasting more than 1 second).