Strictly speaking, a recompile isn't really a performance tuning problem. In fact, a lot of time, recompiles are desired because they reflect changes in statistics which are likely to need a new plan. However, you can get really excessive levels of recompiles because of a variety of different issues. So, identifying the causes can be a pain. Here's one way to get it done. Extended Events for Recompile Causes You knew I was going there. Seriously though, we know that, since SQL Server 2005, all recompiles are at the statement level (unless you call for a recompile of a procedure or create that proc with a RECOMPILE hint). So, capturing the recompiles and getting grotty on the details means capturing statements. Understand that this can be expensive, especially in terms…
One of the many advantages of SQL Cruise is the ability to have enough time during a presentation to be able to answer questions from the people there in great detail. One question came up while I was showing the new functionality of Query Store (available soon in SQL Server 2016, available right now in Azure SQL Database). What happens to plan forcing when you have OPTION RECOMPILE on a query? Great question. I have a favorite procedure I use to illustrate the functionality of parameter sniffing: ALTER PROC dbo.AddressByCity @City NVARCHAR(30) AS SELECT a.AddressID, a.AddressLine1, a.AddressLine2, a.City, sp.Name AS StateProvinceName, a.PostalCode FROM Person.Address AS a JOIN Person.StateProvince AS sp ON a.StateProvinceID = sp.StateProvinceID WHERE a.City = @City; If this procedure is called with the value of 'Mentor' you get…
When faced with a procedure that looks like this: CREATE PROCEDURE dbo.TestProc (@TestValue INT) AS BEGIN IF @TestValue = 1 BEGIN SELECT * FROM Sales.SalesOrderHeader AS soh JOIN Sales.SalesOrderDetail AS sod ON soh.SalesOrderID = sod.SalesOrderID WHERE soh.SalesOrderID = @TestValue END ELSE BEGIN SELECT * FROM Production.Product AS p JOIN Production.ProductDocument AS pd ON p.ProductID = pd.ProductID WHERE p.ProductID = @TestValue END END I used to suggest creating a wrapper procedure in order to avoid the recompiles that occur when the different paths through the IF statement are taken by the optimizer. I mentioned that recently on a post over at SQL Server Central. Gail Shaw (blog | twitter) asked me why I thought there would be a recompile. She said that the optimizer took the query as a whole and…
I saw this question and my immediate thought was "Well, duh, the execution plan is recreated by a recompile." But, someone a bit smarter than myself suggested that, in fact, no plan was stored in cache, forcing a new plan to be created each time the query was run. So, which is it? Does a plan get added to cache and then removed every time the procedure is called, or do you get nothing in cache and the "recompile" is actually a compile every time? As Mr. Owl says, let's find out. I have a small script that looks like this: CREATE TABLE [dbo].[Test]( [col] [varchar] (10) NULL ); CREATE TABLE [dbo].[Test2] ( [col] VARCHAR(10) NULL ); INSERT INTO dbo.Test (col) VALUES ('Val1'), ('Val2'), Â ('Val3') ; INSERT INTO dbo.Test2…
One little potential for performance problems that we've seen comes out of how nHibernate generates it's parameratized queries. It limits the length of any parameter to the length of the column, but if the length of that parameter is less than the column, it uses tha smaller length when declaring the variable. This results in a query that looks like this: exec sp_executesql N'INSERT INTO dbo.users (Name, Password, EmailAddress, LastLogon, LogonId) VALUES (@p0, @p1, @p2, @p3, @p4)',N'@p0 nvarchar(8),@p1 nvarchar(6),@p2 nvarchar(12),@p3 datetime,@p4 nvarchar(8)',@p0=N'Ted Cool',@p1=N'abc123',@p2=N'ted@cool.com',@p3='2008-04-29 14:10:44:000',@p4=N'ted_cool' Note the parameter @p4 which is the parameter mapping to the primary key for the little sample table. In this query it's declared as nvarchar( 8 ) because 'ted_cool' is composed of eight characters. But if we changed it to 'ted_coolish': exec sp_executesql N'INSERT INTO dbo.users…
I live for questions and recently, I had someone ask me, does using SELECT * affect Query Store. My immediate gut reaction was, hell no. Of course it doesn't. Then, yeah, I started thinking. It might. Let's test it and see. The Setup The trick here is to get a good setup. I need a query that's... interesting. Meaning, I need the query to do more than just SELECT * from a table. Although, let's start there: SELECT * FROM dbo.Customers AS c; GO 20 I ran it several times so that it will, for sure, get captured by Query Store (when on Auto, in 2022, one execution of a query may not be enough to see it captured by QS). Then, I took a look at the plan: SELECT…
Recently, a person asked about the costs differences in an execution plan, referencing them as if they were performance measures. The key to understanding performance is to check every metric. When it comes to execution plans, I'm sure I've said this before, so please allow me to repeat myself. The cost numbers shown in an execution plan, which, barring a recompile, will be the same for an execution plan or an execution plan with runtime metrics (aka, estimated and actual plans), are not measures of performance. They do not represent actual metrics. Instead, they are calculations of a theoretical actual performance measurement. So, you can't look at two plans, with two costs, and say, "this plan will perform better." Instead, you can say, "this plan has a lower estimated cost."…
If you're like me, your palms went into a cold sweat to get your grubby fingers on the bits for SQL Server 2022 a few nanoseconds after you heard the announcement. Well, you might be waiting a while. As will I. However, I really do want to start experimenting with some of the functionality right now. If you do to, then join me at Azure SQL Database. Azure SQL Database as a Preview for SQL Server 2022 Yeah, I know it's not the same. I get it. You have a need to run this stuff on big iron, not virtual, and certainly not in the cloud. Still, you want to learn, right? Tons of the new functionality is either already in Azure or incorporates Azure functionality. You can get started…
A question that came up recently was how to track the query compile time. It's actually a pretty interesting question because, there aren't that many ways to tell how long it took to compile the query, and they don't necessarily agree. For most of us, most of the time, compile time for a given query doesn't matter. However, I love telling the story of the query I had on an old system that could run in 90ms, but took 5 minutes to compile. In short, sometimes compile time matters. How To See Query Compile Time If you want to see how long it takes a query to compile, you have, to my knowledge, three options. The first, and possibly easiest, is to look at the plan properties on an execution…
Knowledge of how your system behaves is vital to better control, maintain, and grow the system. While Azure provides all sorts of wonderful assistance within Azure SQL Database, you're still going to need that same knowledge. When it comes to getting detailed information about Azure SQL Database, the tools are a little more limited than with an on-premises instance of SQL Server, or any virtual instance of SQL Server. There are no trace events. To see individual query calls, recompile events, query store behaviors, and so much more, you're going to have to use Extended Events. I'm going to write a series of posts on using Extended Events with Azure SQL Database. Tradition would call for this first post to be an initial how-to. Instead, I want to take a…