Performance Basics: Filter date columns

Troubleshooting performance issues
Another of the basic problems we often find in SQL Server instances with slow queries is because bad filter on date fields.
A lot of times, user complain about high resource utilization on the instance or slow response times, being most of the times, not configuration issues, but bad designed queries).

One of the most common causes is not using a search argument (or SARGs) and therefore not using a full benefit of an index, if you have an index created of course!

Database engine cannot do miracles by itself, you must be careful on what you ask it to do.
So let’s take a look of different approach for the same query: Orders from a year ago (assume getdate() as Jun 01, 2015, because I don’t have data for last year on my test db)

First option: Make the function in the table field, you can see that you use an index scan, in the details, it reads 31K+ rows to obtain only 34

Don't do this, don't be that guy!
 

Second option: make the function in the argument, now you can see that the query plan changes and an index seek is used this time, having full benefit of the date index and retrieving only the 34 rows needed (CPU and IO usage is much lower than the previous option).

An index seek is used now

CPU usage improvement

Ok, that is nice, we can see that we have to make all the function and logic stuff from the parameter side, in order to have a good query, but unfortunately, even If your query is optimal, it will be useless if you don’t have the correct index created.

In this example, I have disabled the date index and executed the last query: you can see that the execution plan changed and for this particular query, the CPU usage is bad, even when the query is optimized.

IDE gives you an alert of a missing index


Even when for these 3 examples they perform similar in a test machine, on large environments having tables with millions of rows and high concurrency, these IO and CPU differences can make a huge difference.

source:

https://docs.microsoft.com/en-us/sql/t-sql/data-types/datetime-transact-sql

 

Comments

Post a Comment

Popular posts from this blog

Checking Azure SQL DB service tier via T-SQL

Image
If you have to manage or work with an Azure SQL database and want to know what service tier the database is currently operating, but you don't have access to the Azure subscription or CLI. Or you want to know the status of the service tier after a scale up or scale down, directly from the database; you can do it via T-SQL Just query the sys.database_service_objectives DMO to obtain this information, this will give information about the service tier, and also will tell you if the database is part of an elastic pool or not. Basic usage (in the context of the database you need the information): SELECT * FROM sys.database_service_objectives; This will return the following information for the current database: I am running on a General Purpose tier with 1 VCore What if you want the information for all the databases created? Just change the context to the master database and execute the following query: SELECT D.name AS DB , D.create_date , SO.edition , SO.service_o
Read more

Quick tip: Zoom in Azure Data Studio

Image
If you finally have given a try t o Azure Data Studio , and if you use it on a regular basis, maybe you want to customize it to suit your needs. Among the huge customization options it has, you can control the text size in form of zoom. To do change it, just use the following keyboard combinations: (Ctrl + = ) For Zoom in. (Ctrl + - ) For Zoom out. (Ctrl + 0) For Zoom reset.  You can see it in action: Also, as with everything on this tool, you can access this functionality from command palette (Ctrl + Shift + P) , and type "zoom" (you can fine-control the editor zoom from here): If you haven't tried it yet, you can download Azure Data Studio here .
Read more

Embed Azure Data Studio Notebooks in your website

Image
Notebooks are a functionality available in Azure Data Studio, that allows you to create and share documents that may contain text, code, images, and query results. These documents are helpful to be able to share database insights and create runbooks that you can share easily. Are you new to notebooks? don't know what are the uses for it? want to know how to create your first notebook? then you can get started in ADS notebooks checking my article for MSSQLTips.com here . Once you have created your first notebooks and share them among your team, maybe you want to share it on your website or blog for public access. even when you can share the file for download, you can also embed it on the HTML code. On this post, I will show you how to do it. What do you need? We will use an online HTML converter, nbviewer , provided by Jypiter website, on that homepage, you just have to provide the link of your .ipynb file (my GitHub Notebook repository for this example). It looks
Read more