Let’s start with the basics. MSDN says “Trace flags are used to temporarily set specific server characteristics or to switch off a particular behavior. Trace flags are frequently used to diagnose performance issues or to debug stored procedures or complex computer systems.”

There are two types of trace flags in SQL Server.

• Session
• Global

We can enable trace flag for the context of a specific query that we can run under a session. Session trace flags can be enabled for a connection and are visible only to that connection. This type of trace flag never affects another session running on your database server and the effect of the trace flag is lost when that connection logs out.

Global trace flags are set at the server level and must be enabled globally. If you will try to enable global trace flag for a particular session then there will not be any impact of this trace flag. Let’s go ahead and understand how we can check the status of trace flags running on your SQL Server instance.

Check Trace Flag Status

We use DBCC TRACESTATUS command to check all running or active trace flags on database system. We can also check the status of a specific trace flag using this DBCC command. The output of this DBCC command has four columns. First column is the trace flag number, second column shows the status of trace flag, third and fourth columns are showing whether the trace flag is enabled at the global level or session level.

The status column of this DBCC command has two values. The value shows whether the trace flag is ON (1) or OFF (0). You can see the below screenshot to understand the trace flags and their status.

DBCC TRACESTATUS

We can run above DBCC command in different ways to get distinct outputs about trace flags status. Here, I am giving you few examples to use DBCC TRACESTATUS command.

Run below command to display the status of all trace flags that are currently enabled globally.

DBCC TRACESTATUS(-1);  
GO

We can even get status of multiple specific trace flags by executing below command. Run below command to get the status of trace flags 8690 and 1118.

DBCC TRACESTATUS (8690, 1118);  
GO

If you want to list all the trace flags that are enabled for the current session then you can get it by running below command.

DBCC TRACESTATUS();  
GO

If you want to check whether trace flag status for specific trace is enabled globally or not then you can run below command.

DBCC TRACESTATUS(8690, -1)

Enable Trace Flag

We enable trace flags to gather more information about certain logs during some transactions. We use DBCC TRACEON command to enable SQL Server trace flags. Here, I will show you different uses of DBCC TRACEON command to enable trace flags.

Remember, all trace flags enabled using DBCC TRACEON command will be disabled once you restart the SQL Server service. If you want to enable any trace flag permanently then we recommend that you should enable them at startup, by using the -T command line option. This ensures the trace flag remains active after a server restart. You can launch SQL Server Configuration Manager and open the Properties window of SQL Server service. Here you can click on Startup Parameters tab. Here you can add your trace flag number along with -T and then click on Ok button to close the window.

If we need to enable a trace flag globally then we need to run below command with -1 option as shown in below screenshot.

DBCC TRACEON (8690, -1)

Now we can check the trace flag status whether it’s been enabled globally on our database system or not.

If you want this trace to be enabled for current connection only then you should run below command. You can remove -1 from above command.

DBCC TRACEON (8690);  
GO

Even you can enable multiple trace flags by running single command. You can run below command to enable trace flag 8690 and 1117 globally. If you want to enable it for current connection only then you can remove -1 from below command.

DBCC TRACEON (8690, 1117, -1);  
GO

Run below DBCC command to check the status of the trace flag that we have enabled by running above command.

DBCC TRACESTATUS

You can see trace flag 8690 is showing along with its value set as 1 for session as well as global in below screenshot.

If you want to check trace flag status for specific trace number then you can run below command. Below command will let you know whether trace flag 8690 is enabled globally or not.

DBCC TRACESTATUS(8690, -1)

Disable Trace Flag

Similarly, we can disable trace flag 8690 that we have enabled in above section by executing another DBCC command known as DBCC TRACEOFF. Run below command to disable trace flag. If we will use -1 value along with DBCC TRACEOFF command that means it will disable this trace flag globally.

DBCC TRACEOFF(8690, -1)

If you want to disable it for current connection only then we will remove -1 from above command and run below command.

DBCC TRACEOFF(8690);   
GO

Now we can again check the status of trace flag 8690 whether this trace is still active or it has been deactivated by above command.

DBCC TRACESTATUS

We can see trace 8690 is disabled now and this trace flag cannot be active in below image.

If you want to check trace flag status for a individual trace then you can run below command.

DBCC TRACESTATUS(8690, -1)

I hope you like this article. Please follow our Facebook page and Twitter handle to get latest updates.

Read More:

• Find Total no of Running Traces files on your Database Server
• Understanding Database Shrink Operation
• How to Avoid Page Split in SQL Server?

The post How to Enable, Disable or Check SQL Server Trace Flags? appeared first on Techyaz.com.

1. Get log space size
2. View wait statistics data
3. Clear wait statistics data
4. Clear latch statistics data

Let’s discuss each of above tasks as use cases of DBCC SQLPERF command.

Read More

• How to get total no of VLFs (Virtual Log Files) inside Log file?
• Understanding Data Files, Pages & Extents
• How to Avoid Page Split in SQL Server?

Get Transaction Log space using DBCC SQLPERF

We can get transaction log space utilization details like how large your transaction logs are and what percentage of them is in use by executing dbcc sqlperf command. Run below command to get the log space details of all database hosted your SQL Server Instance.

--Get Transaction Log Size Details
DBCC SQLPERF(logspace)

You can see the below screenshot; all the columns are self-explanatory. We can see percent of log space used out of total space allocated to the log file.

View and Clear Wait Statistics Data

We can get wait statistics data using dbcc sqlperf command as well. We used to get these details using dbcc sqlperf command in older version of SQL Server instances. Although, you can still get wait statistics data using this command in current versions of SQL Server but now there is a better way to get such details using DMV sys.dm_os_wait_stats.  Here, I will show you wait stats data using DMV only. Run below commands to get wait statistics details of your SQL Server instance.

--Get wait Statistics using DBCC command.
DBCC SQLPERF(waitstats)

--Get Wait Statistics using DMV sys.dm_os_wait_stats
SELECT * FROM sys.dm_os_wait_stats
Order by wait_time_ms desc

You can see the details about all wait types on your SQL Server instance. You can also see the total wait time in milliseconds. These information is very important to deal with performance issues but details like total wait time, total waiting tasks count are not the real-time values. These values are accumulated since last SQL Server service start or last time wait stats was reset using dbcc sqlperf command. It is highly recommended to not clear wait stats data until it is required to do that otherwise you will lose valuable information about wait statistics. Read attached article about why should you not clear wait statistics data stored in DMV sys.dm_os_wait_stats.

If you want to get the correct waiting task counts or waiting time then you need to clear the wait statistics data using dbcc sqlperf command. Here, I will show you the difference after clearing the wait stats. Run below command to clear all wait stats data on your SQL Server Instance.

DBCC SQLPERF(“sys.dm_os_wait_stats”, CLEAR)

You can see I have successfully ran this command in below screenshot.

Now we will again run same command to get the wait statistics data. I ran it and I get below data that is showing different than the one showing before clearing wait stats. This way you can better analyze your database performance and waiting tasks and time to reach on any conclusion.

Clear Latch Statistics Data

I described about clearing wait stats data in above section. Here, I will show you how to clear latch statistics data. Latches are locks for RAM memory structures to prevent concurrent access and ensure serial execution of kernel code. You can run SQL command to check current latch statistics data.

--Get latch statistics data
SELECT * FROM sys.dm_os_latch_stats
ORDER BY wait_time_ms desc

You can see current latch statistics details that is also not showing only real time value but it is accumulated data since last SQL Server service restarted or last time latch wait statistics was reset/cleared.

Now we will clear these latch statistics data to get real time value of latch details. Run below dbcc sqlperf command to clear the latch statistics.

DBCC SQLPERF(“sys.dm_os_latch_stats”, CLEAR)

You can see we have successfully executed above command on our SQL Server Instance.

Now go ahead and run same SQL command to get the latch statistics again. You can see now there is no waiting requests on our SQL Server Instance and you are getting real time value about latches.

Related Articles

• Should We Clear Wait Stats Data from DMV sys.dm_os_wait_stats?
• How to Fix BACKUPBUFFER Wait Type?
• How to Improve Bulk Data Load Performance?

These are the 4 use cases of DBCC SQLPERF command that we use during different database activities. I hope you like this article. Please follow our Facebook page and Twitter handle to get latest updates.

The post 4 Different Usages of DBCC SQLPERF in SQL Server appeared first on Techyaz.com.

Should We Clear Wait Stats Data in SQL Server?

We do not recommend to clear the wait stats data frequently because all valuable information about wait statistics will be vanished if you clear them. Always try to capture wait statistics information before clearing wait stats. This way you would have both before and after data. You can use sys.dm_os_wait_stats DMV to get the SQL Server wait statistics information.

Generally, we reset or clear wait stats information to capture actual data during a given time duration because this DMV does not show only current data. The DMV sys.dm_os_wait_stats accumulates all the counts and wait times since last time SQL Server started or its stats got cleared by dbcc sqlperf command.

Here I will show you how to clear the SQL Server wait statistics and the difference between data captured before clearing the wait stats and after clearing the wait stats.

Get Wait Statistics Data

The DMV sys.dm_os_wait_stats provides essential metrics for diagnosing SQL Server performance problems. This DMV captures all the details like waiting requests counts, signal waits and waiting time since last time SQL Server instance got started or last time you have cleared your wait stats using DBCC SQLPERF command. Run below command to get performance wait stats details of your SQL Server instance.

--Get Wait Statistics using DMV sys.dm_os_wait_stats
SELECT * FROM sys.dm_os_wait_stats
Order by wait_time_ms desc

You can see the details about all the wait types on your SQL Server instance. You can also see the total wait time in milliseconds. This data is very important if you are dealing with any performance issue but as i discussed above details like total wait time, total waiting tasks counts are the accumulated/incremental values.

Reset or CLEAR Wait Statistics

Now you have captured wait stats information since your SQL Server was started. If you want to get only current waiting task counts or waiting time then you need to reset or clear wait statistics data using dbcc sqlperf command. Run below command to clear all wait stats data on your SQL Server Instance.

DBCC SQLPERF(“sys.dm_os_wait_stats”, CLEAR)

You can see I have successfully executed this command in below screenshot.

Now we will again run same command to get the wait statistics data. I ran it and I get below information that is showing different than the one showing before clearing wait stats.

You can see how you will lose valuable information if you will clear your wait stats without capture its earlier data. This way you can also better analyze your database wait types, waiting tasks to reach on any conclusion to fix the issue.

Related Articles

• Understanding SQL Server Wait types
• Identify Disk Bottleneck in SQL Server using Perfmon Disk Counters
• Understanding Resource_Semaphore wait type

Now you understand that we should not reset or clear wait stats frequently because wait stats data is very crucial to deal with any performance issue. You can reset or clear the wait stats but make sure to capture the data before clearing the wait stats. I hope you like this article. Please follow our Facebook page and Twitter handle to get latest updates.

The post Should We Clear Wait Statistics Data from DMV sys.dm_os_wait_stats in SQL Server? appeared first on Techyaz.com.

Let’s check the row counts of tables using script that are using sys.partitions system catalog view and compare the output of a table with the output of SELECT COUNT (*) .

Get Row Counts of All Tables using sys.partitions or DMV sys.dm_db_partition_stats

System catalog view sys.partitions contains a row for each partition of all the tables and most types of indexes in the database. All tables and indexes in SQL Server contain at least one partition, whether they are explicitly partitioned. Run below script to get the row count of all tables in a database. If you want to find row counts using DMV sys.dm_db_partition_stats, i would suggest you to visit our attached article how to find row counts of all tables in a database.

--Run below script to get the row counts of all tables in a database.
SELECT      SCHEMA_NAME(O.schema_id) + '.' +
        O.Name As [Schema.TableName], SUM(P.rows) AS 'Total_RowCount'
FROM        sys.objects O
INNER JOIN sys.partitions P ON O.object_id = P.object_id
WHERE     O.type = 'U'
GROUP BY    O.schema_id, O.Name

You can see row count of all tables in below screenshot.

Now we will check row count of a table using SELECT COUNT(*) command. I ran below command and you can see that row count is not same here. Row count came using sys.partitions system catalog view has more value than the row count of SELECT COUNT (*).

As i said above “sys.partitions or DMV sys.dm_db_partition_stats contains a row for each partition of all the tables and most types of indexes in the database” so when you are fetching row counts of tables using sys.partitions system catalog view or DMV sys.dm_db_partition_stats, it will display the row counts after adding indexes row counts and table row counts. That is why value is showing more in sys.partitions or in sys.dm_db_partition_stats output.

Suppose you have a table with 5000 rows in it and you’ve defined 5 indexes on it, the above query would show 5*5000=25000 rows for that table. So, it isn’t just that some rows are inconsistence, it is that some rows are multiplied by the number of indexes they have.

If we need correct row count using the system cataog view sys.partitions then we need to put a condition in WHERE clause for index_id column. We can also get total row count of all the heap tables or for all the clustered index tables as well by changing the value of index_id mentioned in WHERE clause. Index_id value indicates the ID of the index within the object to which this table belongs.

• 0 = heap
• 1 = clustered index
• 2 or greater = nonclustered index

We can pass index_id value in above script to get the total row count of all tables whether it is heap or clustered index table. If you need total row count of all the tables present in a database, you should put index_id<2 condition in WHERE clause. Run below scripts to get the row count of all heap & clustered index tables.

--Run below script to get the row counts of all tables in a database.
--Change the value of index_id to 0 or 1 to get total row count to heap or cluster index tables.
SELECT      SCHEMA_NAME(O.schema_id) + '.' +
        O.Name As [Schema.TableName], SUM(P.rows) AS 'Total_RowCount'
FROM        sys.objects O
INNER JOIN sys.partitions P ON O.object_id = P.object_id
WHERE     O.type = 'U'
		AND P.index_id<2  
GROUP BY    O.schema_id, O.Name

Now you can run above script and compare the output with the SELECT COUNT(*) output. We can see row count for given table is showing same after adding idex_id condition in below screenshot.

I hope you like this article. You can comment us about your questions in below section. Please follow us on our facebook page and on Twitter handle to get latest updates.

The post Why is table row count showing incorrect value in sys.partitions or in DMV sys.dm_db_partition_stats appeared first on Techyaz.com.

• Tables
• Heap tables only
• Clustered index tables only
• Tables having specific Schema

We can get total row counts of all tables using system catalog view sys.partitions or DMV sys.dm_db_partition_stats. Read attached article if you are getting wrong row counts using this DMV or catalog view. Let’s check the row counts using both ways in this tip.

Get Row Counts of Tables using sys.partitions

System catalog view sys.partitions contains a row for each partition of all the tables and most types of indexes in the database. Special index types such as Full-Text, Spatial, and XML are not included in this view. All tables and indexes in SQL Server contain at least one partition, whether or not they are explicitly partitioned. Run below script to get the row count of all tables in a database.

--Run below script to get the row counts of all tables in a database.
SELECT      SCHEMA_NAME(O.schema_id) + '.' +SELECT      SCHEMA_NAME(O.schema_id) + '.' + O.name As [Schema.TableName],  SUM(P.rows) AS [Total_RowCount] 
FROM        sys.objects O
INNER JOIN sys.partitions P ON O.object_id = P.object_id
WHERE     O.type = 'U' AND P.index_id<2  
GROUP BY    O.schema_id, O.name
ORDER BY [Total_RowCount] desc

You can see row count of all tables in below screenshot.

We can get total row count of all the heap tables or for all the clustered index tables as well by changing the value of index_id mentioned in WHERE clause. Index_id value indicates the ID of the index within the object/table. If index_id value is 0 that means that table doesn’t have any cluster index and a Heap table.

• 0 = heap
• 1 = clustered index
• 2 or greater = nonclustered index

We can pass index_id value in above script to get the total row count of all tables whether it is heap or clustered index table. Run below scripts to get the row count of all heap tables.

--Run below script to get the row counts of all Heap (tables without cluster index) in a database.
--Change the value of index_id from 0 to 1 to get total row count to cluster index tables.
SELECT      SCHEMA_NAME(O.schema_id) + '.' +
        O.name As [Schema.TableName], SUM(P.rows) AS [Total_RowCount]
FROM        sys.objects O
INNER JOIN sys.partitions P ON O.object_id = P.object_id
WHERE     O.type = 'U'
		AND P.index_id=0  --Change this value according to your need
GROUP BY    O.schema_id, O.name
ORDER BY [Total_RowCount] desc

You can see there are only two tables that have no cluster index in above screenshot. We can also get the row count of all tables having a specific schema by adding a condition in WHERE clause.

--Run below script to get the row count of all tables having dbo Schema in a database.
SELECT      SCHEMA_NAME(O.schema_id) + '.' +
        O.name As [Schema.TableName], SUM(P.rows) AS [Total_RowCount]
FROM        sys.objects O
INNER JOIN sys.partitions P ON O.object_id = P.object_id
WHERE     O.type = 'U'
		AND P.index_id<2  
		AND SCHEMA_NAME(O.schema_id) ='dbo'--Change dbo with your schema name for which you need row counts
GROUP BY    O.schema_id, O.name
ORDER BY [Total_RowCount] desc

You can see the output in below image. You can make changes as per your need.

Get Row Count of All Tables using DMV sys.dm_db_partition_stats

DMV sys.dm_db_partition_stats returns page and row-count information for every partition in the current database. We will leverage this DMV to get row count details of all tables. Run below script to get the row count of all tables using this DMV.

--Run below script to get the row count of all tables in a database.
SELECT QUOTENAME(SCHEMA_NAME(O.schema_id)) + '.' + QUOTENAME(O.name) AS [Schema.TableName]
      , SUM(p_dmv.row_count) AS [Total_RowCount]
FROM sys.objects AS O
      INNER JOIN sys.dm_db_partition_stats AS p_dmv
            ON O.object_id = p_dmv.object_id
WHERE  O.type = 'U'
        AND p_dmv.index_id < 2
GROUP BY
      O.schema_id
      , O.name
ORDER BY [Schema.TableName]
GO 

We can see the output of above script in below image.

We can pass index_id value in above script to get the total row count of all tables whether it is heap or clustered index table. Run below scripts to get the row count of all heap & clustered index tables.

--Run below script to get the row counts of all Heap (tables without cluster index) in a database.
--Change the value of index_id from 0 to 1 to get total row count to cluster index tables.
--Change the schema name if you want the row count of tables that are having specific schema.
SELECT QUOTENAME(SCHEMA_NAME(O.schema_id)) + '.' + QUOTENAME(O.name) AS [Schema.TableName]
      , SUM(p_dmv.row_count) AS [Total_RowCount]
FROM sys.objects AS O
      INNER JOIN sys.dm_db_partition_stats AS p_dmv
            ON O.object_id = p_dmv.object_id
WHERE  O.type = 'U'
        AND p_dmv.index_id < 2
	AND SCHEMA_NAME(O.schema_id) = 'dbo'--Change the name of your Schema to get row count of all tables having this schema.
GROUP BY
      O.schema_id
      , O.name
ORDER BY [Schema.TableName]
GO 

And here is the output:

I hope you like this article. You can comment about your questions in comment section. Please follow us on our facebook page and on Twitter handle to get latest updates.

Related Articles:

• Why rou count showing incorrect value in sys.partitions?
• Understanding Different Types of SQL JOINs with Examples
• Create a Logon Trigger to Restrict sysadmin logins to Connect to SQL Server during a given time Interval
• How to Protect Stored Procedure Code/Script?

The post How to Get Row Count of All Tables of a Database appeared first on Techyaz.com.