SQL SERVER – Get Query Running in Session

I was recently looking for syntax where I needed a query running in any particular session. I always remembered the syntax and ha d actually written it down before, but somehow it was not coming to mind quickly this time. I searched online and I ended up on my own article written last year SQL SERVER – Get Last Running Query Based on SPID. I felt that I am getting old because I forgot this really simple syntax.

This post is a refresher to me. I knew it was something so familiar since I have used this syntax so many times during my performance tuning project.

Run the following query to find out what the latest query that was executed in the session. There are various methods mentioned in my earlier post, so here I am picking only the one that I use most of the time.

Please use Shortcut – CTRL+T or enable “result to text” in the resultset to get formatted output.

SELECT @sqltext = sql_handle
FROM sys.sysprocesses
WHERE spid = (YourSessionID)

You can find a range of active session IDs in your system by running system stored procedure sp_who2.

The following is the resultset where I have selected the session id that is the same as from where I am running above statement.

sysprocesses SQL SERVER – Get Query Running in Session

Additionally, you can use following T-SQL script as well.

CROSS APPLY sys.dm_exec_sql_text(most_recent_sql_handle)
WHERE session_id = (yoursessionID)

Reference: Pinal Dave (http://blog.SQLAuthority.com)

SQL SERVER – Check the Isolation Level with DBCC useroptions

In recent consultancy project coordinator asked me – “can you tell me what is the isolation level for this database?” I have worked with different isolation levels but have not ever queried database for the same. I quickly looked up bookonline and found out the DBCC command which can give me the same details.

You can run the DBCC UserOptions command on any database to get few details about dateformat, datefirst as well isolation level.

DBCC useroptions

Set Option                  Value
--------------------------- --------------
textsize                    2147483647
language                    us_english
dateformat                  mdy
datefirst                   7
lock_timeout                -1
quoted_identifier           SET
arithabort                  SET
ansi_null_dflt_on           SET
ansi_warnings               SET
ansi_padding                SET
ansi_nulls                  SET
concat_null_yields_null     SET
isolation level             read committed

I thought this was very handy script, which I have not used earlier. Thanks Gary for asking right question.

Note: This optional will only give you isolation for current connection where the command has ran and not at server level.

Reference: Pinal Dave (http://blog.sqlauthority.com)

SQL SERVER – Find Max Worker Count using DMV – 32 Bit and 64 Bit

During several recent training courses, I found it very interesting that Worker Thread is not quite known to everyone despite the fact that it is a very important feature. At some point in the discussion, one of the attendees mentioned that we can double the Worker Thread if we double the CPU (add the same number of CPU that we have on current system). The same discussion has triggered this quick article.

Here is the DMV which can be used to find out Max Worker Count

SELECT max_workers_count
FROM sys.dm_os_sys_info

Let us run the above query on my system and find the results.

maxworkercount SQL SERVER   Find Max Worker Count using DMV   32 Bit and 64 Bit

As my system is 32 bit and I have two CPU, the Max Worker Count is displayed as 512.

To address the previous discussion, adding more CPU does not necessarily double the Worker Count. In fact, the logic behind this simple principle is as follows:

For x86 (32-bit) upto 4 logical processors  max worker threads = 256
For x86 (32-bit) more than 4 logical processors  max worker threads = 256 + ((# Procs – 4) * 8)
For x64 (64-bit) upto 4 logical processors  max worker threads = 512
For x64 (64-bit) more than 4 logical processors  max worker threads = 512+ ((# Procs – 4) * 16)

In addition to this, you can configure the Max Worker Thread by using SSMS.

Go to Server Node >> Right Click and Select Property >> Select Process and modify setting under Worker Threads.

maxworkersetting SQL SERVER   Find Max Worker Count using DMV   32 Bit and 64 Bit

According to Book On Line, the default Worker Thread settings are appropriate for most of the systems.

Reference: Pinal Dave (http://blog.SQLAuthority.com)

SQL SERVER – Plan Recompilation and Reduce Recompilation – Performance Tuning

Recompilation process is same as compilation and degrades server performance. In SQL Server 2000 and earlier versions, this was a serious issue but in SQL server 2005, the severity of this issue has been significantly reduced by introducing a new feature called Statement-level recompilation. When SQL Server 2005 recompiles stored procedures, only the statement that causes recompilation is compiled, rather than the entire procedure. Recompilation occurs because of following reason:

  • On schema change of objects.
    • Adding or dropping column to/from a table or view
    • Adding or dropping constraints, defaults, or rules to or from a table.
    • Adding or dropping an index to a table or indexed view if index is used by the plan.
    • Adding or dropping trigger from a table
    • Dropping statistics from a table that is used in plan.
  • On change of the SET options: When a compiled plan is created, the SQL server also stores the environmental setting of a connection (SET option) with it. If the same stored procedure is executed by another connection that has a different SET option, then the existing cached plan is not reused. To reduce recompilation caused by Environment change, we should not change the SET options in the connection setting and stored procedure.
  • On statistics change of tables: Every time the SQL Server uses an already cached compiled plan, it checks the optimality before using it. SQL Server decides whether the plan is optimal for current amount of data in underlying tables. It could be that the data amount in underlying tables changed so much that the previously created plan is not optimized. For each table & index, the SQL server maintains a modification counter and if the counter values exceed the defined threshold, the previously created compiled plan is considered stale plan and a new plan is created.

Detecting recompilations: The below query retrieves the top 10 statements for which the recompilation count is maximum. Here, plan_generation_num returns a number that indicates the recompilation count of a statement.

DB_NAME(st.dbid) AS DbName,
sys.dm_exec_query_stats qs
CROSS APPLY sys.dm_exec_sql_text(sql_handle) AS st
ORDER BY plan_generation_num DESC

Another tool is to get recompilation details is the SQL Profiler. Along with the statements text, it also tells the reason of recompilation in EventClass and EventSubClass columns.

Reducing recompilations: We have no option to avoid “schema change” and “SET options” based recompilation. But we have following query and stored procedure hints to avoid “statistics change” based recompilation:

  • KEEP PLAN hint: The modification counter threshold for a temporary table is 6. This implies that when a stored procedure that creates a temporary table inserts 6 or more rows into this table, Stored Procedures will be recompiled as soon as this table is accessed. For permanent tables, this threshold is at least 500. We can increase the first threshold for the temporary table (6) to same as that of the permanent table (500) by using the KEEP PLAN query hint in the statement where the temporary table is used. For example,

SELECT TT.col4, SUM(PermTable.col1)
FROM dbo.PermTable
ON PermTable.col1 = TT.col2

  • KEEPFIXED PLAN hint: This hint completely avoids “statistics change” based recompilation of a query.

SELECT col1, col2
FROM dbo.PermTable
WHERE col3 < 100

  • Use Table variable instead of temporary tables: Because a change in cardinality of a table variable does not cause recompilations, consider using a table variable instead of a temporary table when faced with a problem of excessive recompilation. However, the use of table variables can lead to poorer query plans. Distribution statistics are not stored for table variables, and cardinality is only available during recompilation and not during initial compilation. One has to check whether this is the case and make an appropriate trade-off.

Reference: Pinal Dave (http://blog.SQLAuthority.com)

SQL SERVER – Find the Size of Database File – Find the Size of Log File

I encountered the situation recently where I needed to find the size of the log file. When I tried to find the script by using Search@SQLAuthority.com I was not able to find the script at all. Here is the script, if you remove the WHERE condition you will find the result for all the databases.

SELECT DB_NAME(database_id) AS DatabaseName,
Name AS Logical_Name,
Physical_Name, (size*8)/1024 SizeMB
FROM sys.master_files
WHERE DB_NAME(database_id) = 'AdventureWorks'

Reference: Pinal Dave (http://blog.SQLAuthority.com)

SQL SERVER – Get the List of Object Dependencies – sp_depends and information_schema.routines and sys.dm_sql_referencing_entities

Recently, I read a question on my friend Ritesh Shah‘s SQL site regarding the following: sp_depends does not give appropriate results whereas information_schema.routines does give proper answer.

I have quite often seen that information_schema.routines gives proper dependency relationship where assp_depends returns an incorrect answer. However, as per book online sp_depends will be deprecated, and instead, sys.dm_sql_referencing_entities and sys.dm_sql_referenced_entities are recommended.

Let us quickly see where sp_depends fail and other solutions work fine.

Let us first create two scenarios.

Scenario 1: Normal Table Creation Order, where objects are created first and then used afterwords.

CREATE TABLE dbo.TestTable
Name VARCHAR(100))
-- dbo.First is not created yet
ID, Name
FROM TestTable

Scenario 2: Objects are created afterwords and they are referenced first.

CREATE TABLE dbo.TestTable
Name VARCHAR(100))
ID, Name
FROM TestTable
-- dbo.First is already created

Now let us run following three queries on both the scenarios.

-- Method 1: Using sp_depends
sp_depends 'dbo.First'
-- Method 2: Using information_schema.routines
FROM information_schema.routines ISR
-- Method 3: Using DMV sys.dm_sql_referencing_entities
SELECT referencing_schema_name, referencing_entity_name,
referencing_id, referencing_class_desc, is_caller_dependent
FROM sys.dm_sql_referencing_entities ('dbo.First', 'OBJECT');

Result from Scenario 1

sp depends1 SQL SERVER   Get the List of Object Dependencies   sp depends and information schema.routines and sys.dm sql referencing entities

Result from Scenario 2

sp depends2 SQL SERVER   Get the List of Object Dependencies   sp depends and information schema.routines and sys.dm sql referencing entities

It is clear that sp_depends does not give proper/correct results when the object creation order is different or following deferred name resolution.

I suggest the use of the third method, in which sys.dm_sql_referencing_entities is used.

Use the following script to get correct dependency:

SELECT referencing_schema_name, referencing_entity_name,
referencing_id, referencing_class_desc, is_caller_dependent
FROM sys.dm_sql_referencing_entities ('YourObject', 'OBJECT');

Let me know the type of scripts you use for finding Object Dependencies. I will post your script with due credit.

Reference: Pinal Dave (http://blog.SQLAuthority.com)

SQL SERVER – Location of Resource Database in SQL Server Editions

While working on project of database backup and recovery, I found out that my client was not aware of the resource database at all. The Resource database is a read-only database that contains all the system objects that are included with SQL Server.

Location of this database is at different places in the different version of SQL Server.

In SQL Server 2008:

<drive>:\Program Files\Microsoft SQL Server\MSSQL10.<instance_name>\Binn\.
The Resource database cannot be moved.

ResourceDB location in SQL Server 2008
ResourceDB location in SQL Server 2008

In SQL Server 2005:

Same place where master database is located.
The Resource database have to move with master database.

You can run following commands to know resourcedb version and last updated time.

SELECT SERVERPROPERTY('ResourceVersion') ResourceVersion,
SERVERPROPERTY('ResourceLastUpdateDateTime') ResourceLastUpdateDateTime

resourcedbver SQL SERVER   Location of Resource Database in SQL Server Editions

Reference: Pinal Dave (http://blog.SQLAuthority.com)

SQL SERVER – Recently Executed T-SQL Query

About a year ago, I wrote blog post about SQL SERVER – 2005 – Last Ran Query – Recently Ran Query.  Since, then I have received many question regarding how this is better than fn_get_sql() or DBCC INPUTBUFFER.

The Short Answer in is both of them will be deprecated.

Please refer to following update query to recently executed T-SQL query on database.

SELECT deqs.last_execution_time AS [Time], dest.TEXT AS [Query]
FROM sys.dm_exec_query_stats AS deqs
CROSS APPLY sys.dm_exec_sql_text(deqs.sql_handle) AS dest
ORDER BY deqs.last_execution_time DESC

Reference : Pinal Dave (http://blog.SQLAuthority.com)

SQL SERVER – Get Query Plan Along with Query Text and Execution Count

Quite often, we need to know how many any particular objects have been executed on our server and what their execution plan is. I use the following handy script, which I use when I need to know the details regarding how many times any query has ran on my server along with its execution plan. You can add an additional WHERE condition if you want to learn about any specific object.

SELECT cp.objtype AS ObjectType,
OBJECT_NAME(st.objectid,st.dbid) AS ObjectName,
cp.usecounts AS ExecutionCount,
st.TEXT AS QueryText,
qp.query_plan AS QueryPlan
FROM sys.dm_exec_cached_plans AS cp
CROSS APPLY sys.dm_exec_query_plan(cp.plan_handle) AS qp
CROSS APPLY sys.dm_exec_sql_text(cp.plan_handle) AS st
--WHERE OBJECT_NAME(st.objectid,st.dbid) = 'YourObjectName'

In result set, you will find the last column QueryPlan with XML text in it. Click on it, and it will open a new tab displaying the graphical execution plan.

explan1 SQL SERVER   Get Query Plan Along with Query Text and Execution Count

You can also save the newly opened tab in XML format with the extension .sqlplan. When this new file is opened in SQL Server Management Studio, it will automatically display the graphical representation of the XML Plan.

explan2 SQL SERVER   Get Query Plan Along with Query Text and Execution Count

It is a quite simple script, but it works all the time.

Reference : Pinal Dave (http://blog.sqlauthority.com)

SQL SERVER – Measure CPU Pressure – CPU Business

Many a time, DBAs face the following question: can the CPU handle the current transaction? I have seen many DBAs getting confused on whether they should get a CPU with higher clock speed or more number of CPUs when it is time to upgrade system. As we all know, the CPU (or the Central Processing Unit) is the core of any system. The CPU is responsible for not only SQL Server operations but also all the OS (Operating System) tasks related to the CPU on the server.

It is quite possible that although we are running very few operations on our SQL Server, we still do not obtain the expected results. This is when the SQL Server CPU has to be checked. Again, please note that I am not suggesting that the CPU is the root cause of poor performance in this case. There are ample chances that the Memory, Input/Output or Tempdb can be responsible for performance degradation. However, to find out the “real” culprit, we must run a test; here, we intend to start running tests on the CPU.

Let us understand the simple meaning of CPU pressure. CPU pressure is a state wherein the CPU is fully occupied with currently assigned tasks and there are more tasks in the queue that have not yet started.

We can run several different types of queries to test the CPU. However, my personal favorite is the following one. The resultset will contain as many rows as those being used by the CPU SQL Server.

FROM sys.dm_os_schedulers
WHERE scheduler_id < 255;

Above query will give result something like following.

cpupress SQL SERVER   Measure CPU Pressure   CPU Business

Before we go ahead, let us familiarize ourselves with the terms. First of all, we should understand what a “scheduler” refers to in the SQL Server. Each SQL Server instance behaves and acts like any OS internally. It has schedules and also it synchronizes concurrent tasks by itself without going to OS. Each instance can handle many concurrent requests. However, there is a point at which the instance cannot accept any more requests. SQL Servers have dynamic management view (DMV), which in fact keeps a log of all the scheduler-related tasks and also it provides a good idea about the CPU pressure.

current_tasks_count is the number of counts of the currently running task. However, we should give utmost consideration to runnable_tasks_count. If this number is higher, it indicates that a large number of queries, which are assigned to the scheduler for processing, are waiting for its turn to run. This gives a clear indication of the CPU pressure. Additionally, count pending_disk_io_count displays the tasks that are yet to be processed in the scheduler. For better processing, this count is expected not to be very high.

When we say that the numbers are high or low, it does not make any sense unless we compare it to a standard or any other known count. Therefore, here, these numbers are compared to the worker counts (current_worker_count). If current_worker_count is 24 and there are 1000 tasks in queue, then this is not a good scenario. Thus, you can always look at the numbers and make your own judgement here.

Well, in this article, I have done a lot of talking with very little script. Please do let me know your opinions regarding whether this was simple enough to convey the message regarding CPU pressure.

Reference : Pinal Dave (http://blog.SQLAuthority.com)