May 142019
 

There is still no love for SQL Server Management Studio. SSMS v18 Beta included a Dark Theme, but somehow that did not make it into the final release of the product. Other apps have this capability (Visual Studio, Visual Studio Code, Chrome, Edge) but not SSMS. If you are like me, sometimes you want to switch your apps to a dark theme for added flair. A quick web search will reveal several ways to hack SSMS to display a dark theme, but that may not produce the best results. So, until Microsoft decides to include it, here is how I have been customizing SSMS for the last several years.

Before:

After:

These colors may not work for everyone, but it does demonstrate how you can customize SSMS to fit your tastes.

To get started, from the menu select Tools and then Options.

In the Options window we’ll need to select Environment and the Fonts and Colors.

For Show settings for, select Text Editor and change the font size to 12.

Under Display items, select Plain Text and then click the Custom button next Item foreground.

In the Color window, change the Red to 228, Green to 228, Blue to 228, and then click OK.

Under Item background, select Black from the dropdown list.

Follow the same steps to change each of these other items below.

Text Editor:

  1. Font size: 12

Display Items:

  1. Plain Text
    1. Foreground: 228,228,228
    2. Background: Black
  2. Selected Text
    1. Background: Yellow
  3. Line Number
    1. Foreground: 255, 255, 128
  4. Comment
    1. Foreground: 255, 175, 45
  5. Keyword
    1. Foreground: 127, 176, 228
  6. SQL Operator
    1. Foreground: 199, 199, 199
  7. SQL Stored Procedure
    1. Foreground: 128, 255, 128
  8. SQL String
    1. Foreground: Yellow
  9. SQL System Function
    1. Foreground: 220, 120, 255
  10. SQL System Table
    1. Foreground: 128, 255, 128
  11. String
    1. Foreground: Yellow
  12. URL Hyperlink
    1. Foreground: Cyan

Next, select Grid Results from the Show Settings For dropdown list. Change the font to Microsoft San Serif and the size to 12.

Then select Text Results from the Show Settings For dropdown list and change the font to Consolas and the font size to 12.

Changing the Grid or Text Results will require you to close and reopen SSMS for the changes to take effect. The last thing is to display the line numbers. On the left side of the Options window, select Text Editor, Transact-SQL, and then General. Click the check box next to Line numbers and click OK.

Once you have the colors you want, then go to Tools and click Import and Export Settings.

This wizard allows you to save all the customization settings to a settings file. The file can be used as a backup, or you can use that file to import to SSMS running on another workstation.

Happy customizing!

Jul 212015
 

Have you ever called Microsoft Customer Support for help with a SQL Server performance issue? The engineer will typically send you a self-extracting exe file with instructions to unzip the file on the server and run the pssdiag.cmd file to collect data while you reproduce your issue. Afterwards, you’re supposed to zip up the contents of the output folder and send it back to the engineer. How did the engineer compile that collection code so quickly? The answer is PSSDiag Configuration Manager, or simply DiagManager.

DiagManager is just a tool that you use to configure which performance items are to be collected by the SQLDIAG utility. The configuration files that come with SQLNexus (see my previous post) can be edited by hand, but sometimes it is much easier to use a GUI, and that’s where DiagManager can help you.

Version 12.0.0.1001 is the latest and to get it installed, just download it from codeplex and run SetupX86.msi. Go to the Start Menu and click on PSSDiag Configuration Manager to launch the program. Initially, you’ll get a warning screen to remind you the SQL client tools are required to run the collector. Just click OK to continue.

The first thing you’ll likely notice is a line of tabs across the top for different versions of SQL Server. Even though there are tabs for SQL Server 7.0 and 2000, they are no longer supported by this tool. If you really need a collection tool for those old version, then follow this link. But on the bright side, this latest version supports up to SQL Server 2014.

To get started using DiagManager, choose your version of SQL Server by selecting a tab.

Next, select the type of processor that your server has; for example, Intel or AMD. Itanium-based processors are no longer support for current versions of SQL Server. My server uses AMD processors, so I’ll need to click the AMD button.

Now you will need to enter the Machine name and the Instance name that will be monitored. The default values “.” and “*” will capture data for all SQL Server instances on the local machine. Typically for troubleshooting you won’t need to capture data for all instances. If your SQL Server installation is running on a Windows Failover Cluster, then you need to enter the SQL Server Virtual Name and not the physical host name. You’ll also need to specify the credentials used to connect to SQL Server. In my example, I’ll be monitoring a stand-alone instance named TRONSQL2012 and connecting using Windows Authentication.

The section below that contains the Machine-wide Diagnostics are the counters you would normally see in Performance Monitor. All of the counters for SQL Server are checked by default as well as several for the Windows OS. At this point, you can add or remove counters that you want to capture. You can also specify the max file size (in MB) as well as the collection interval. The fault values are 256MB and 5 seconds.

Unchecking the Perfmon box will disable the collection of all perfmon counters.

The three check boxes above the perfmon box are for the Windows Event Logs. By default, DiagManager will only grab the event logs when SQLDIAG is shutdown; however, you can check the Startup box and it will grab them at both startup and shutdown. This is helpful if you need to do a before after comparison of the Windows Event logs. The collected data will be saved in the output folder as comma-separated (CSV) files. Unchecking the Event Logs box will disable the collection of the event logs.

In the top right half of the window, you will see Instance-specific Diagnostics; aka SQLDIAG and SQL trace. By default, SQLDIAG will only collect its data at shutdown, but you can check the Startup box to get before and after collections. This works the same as the event log collection. If you want to see the TSQL code that is used to generate this output, then just open the MSDiagProcs.sql file in Management Studio. Unchecking the SQLDIAG box will disable the collection of the SQLDIAG data.

Below SQLDIAG, you will see three tabs: Database Engine, Analysis Services, and XEvent. The Analysis Services tab is disabled for the public version of DiagManager, and in the current version, the XEvent tab should not be used to collect database engine data. This leaves the Database Engine tab, which is controls the events that will be captured using a SQL Trace. If you have ever used SQL Profiler to create a trace, then you should be very familiar with these options. Simply browse through the list to check or uncheck the items you need to collect.

Above the list of trace events, you will see a Trace checkbox and two dropdown lists. Unchecking the Trace box will disable the collection of any trace data. The “Max file size” dropdown just controls how large the trace files will get before they rollover. The default is 350MB. The Template dropdown list is a way to quickly check/uncheck events that will be collected. There are three templates included: General Performance, Details Performance, and PSSReplay. The PSSReplay is similar to the replay template within Profiler. General vs Details events is basically batch-level events vs statement-level events. The coolest thing about the trace templates, is that you can create your own. To do so, navigate to the TraceTemplates folder in the DiagManager install folder (C:Program Files (x86)MicrosoftPssdiagTraceTemplates). Open the folder that matches your SQL version, and make a copy of one of the XML files. Open that cop in your favorite XML editor and just change the “enabled” value to “true” for the events to collect and “false” for the events you do not want to collect, and then save the XML file.

The next time you open up DiagManager you should see your custom template in the dropdown list.

One big word of caution; a SQL Trace will cause the HEAVIEST overhead to your SQL Server, so be careful what items you choose to capture. If you are worried about adding too much overhead to your SQL Server, then consider using the “XEvent Batch_RPC Only” option in the Custom Diagnostics section. Checking this option will create an Extended Events session that will only capture “rpc_completed” and “sql_batch_completed” events. An Extended Events sessions adds far less overhead than a SQL Trace, but DO NOT enable both the SQL Trace box and the XEvent Batch_RPC option. Doing so would run both a trace and xevent session and add a lot of overhead to your SQL Server.

For the other items in the Custom Diagnostics section, you can select each one to see a description of what they do in the Instructions window. For example, the SQL 2012 Perf Stats script will gather data from DMVs about various items such as: top N queries, missing indexes, statistics, and even Availability Group info.

The Custom Diagnostics section is exactly as it sounds. You can create your own TSQL scripts or other types of custom collectors. Navigate to the CustomDiagnostics folder in the DiagManager install folder (C:Program Files (x86)MicrosoftPssdiagCustomDiagnostics), and create a new folder a new folder for your custom code. In my example, I created a folder called “SQL Database List” and added three files to the folder.

The first is just a TSQL script that has one command: “SELECT * FROM sys.databases;”. The DatabaseList.cmd is a Windows batch file that calls sqlcmd.exe and runs the DatabaseList.sql file.

The CustomDiag.XML file was copied from another custom folder and edited to reference the TSQL file and define whether it runs at startup or shutdown.

Once you have your custom collector created, close and open DiagManager and you should see your custom collector in the Custom Diagnostics list.

My example just captures a list of the databases, but you could easily create a script that captures data internal to one of your application databases.

Now that you have finally selected all the items that you want to collect, click the Save button at the top of the window. A Save Configuration window will pop up asking you what and where to save the collector files.

The Filename filed is where you want to save the PSSDiag.XML configuration file as well as all the collection scripts and tools. The option to create a CAB file is where DiagManager will compress and save all the files and scripts from the top folder. The CAB file is just a compressed (ZIP) file, and makes it easier to copy your collection tools to a remote server where the CAB file can then be uncompressed.

The last option available in DiagManager is the ability to compare PSSDiag.XML files. Just click the “Compare Config Files” button at the top of the window.

In the popup window, just select the two XML files that you want to compare, click the Compare button, and the output will be shown below.

Once you have created your collection set and/or the CAB file. All you need to do is copy the CAB file to the server to be monitored and execute the pssdiag.cmd batch file.

It will open a command prompt window and kick off each of the collection items you specified. Once you see the message “SQLDIAG Collection started” in green-colored font, then you can go recreate your performance issue.

Once you are ready to stop the collection, just press Ctrl+C. All of the data collected will be placed in the output folder, including any custom scripts that you may have created. That output folder can then be copied back to your workstation and fed into SQL Nexus for analysis. See my previous post for a detailed tutorial.

DiagManager definitely has a lot of configuration options to choose from, so learning which ones to use might take time. A good way to get started it is turn off all options except for one, and then run the data capture. Then you can look at the results in the output folder to really understand what that option does.

DiagManager, as well as SQLNexus, was created by the support teams at Microsoft, and they have made both of them freely available to the public to help in troubleshooting issues, so there’s no excuse not to try them out. I hope this tutorial will help you get started.

Mar 102015
 

The year is 2015, and I’m still surprised by how many people have never heard of SQL Nexus; although it has been available since approximately 2008. SQL Nexus is a tool that that is designed to collect and analyze data to help you troubleshoot SQL Server performance issues. There are plenty of tools on the market that can easily do this, but this one is free and it can be used to monitor SQL Server 2005 through SQL Server 2014. The tool was originally created by the SQL Server product support team, and is still used today as the primary tool for collecting and analyzing performance issues.

SQL Nexus does not require you to install anything on the server that you intend to monitor, but you will need to copy some data collector files to it. The collector files just run SQLDIAG and some other TSQL scripts to collect data from your monitored server. Below is a list of some of the data that is collected.

  1. Windows Performance Monitor counters
  2. Windows event logs (system, application, and security logs)
  3. Windows configuration settings
  4. SQL Server trace events
  5. SQL Server wait stats
  6. SQL Server blocking info
  7. SQL Server configuration settings
  8. SQL Server dump files

Once the data is collected from the monitored server, you copy all the data files back to another machine for analysis within SQL Nexus.

Installation & Setup
There are a few requirements to setting up SQL Nexus. First, make sure you have the .NET Framework 4.0 installed. Next, SQL Nexus requires an instance of SQL Server to store and analyze the imported data. Ideally, you would put it on a TEST server, but you can also use SQL Server Express Edition running on your workstation. If your workstation that will do the analysis is running SQL Server 2008 or SQL Server 2008 R2, then download and install the Microsoft Report Viewer 2010 control. If your workstation is running SQL Server 2012 or higher, then you can skip this step. Regardless of the version, download and install the Report Viewer hotfix. The hotfix addresses some bugs that will prevent some of the SQL Nexus reports from working properly.

Next, download and install the RML Utilities.

Now that you have all the prerequisites installed, download the latest version of SQL Nexus from the CodePlex website. Just uncompress the Release4.0.0.64.ZIP file and place on your desktop. Within the Release4.0.0.64 folder, you should see a subfolder called DataCollectors, and another compressed file called SqlNexus4.0.0.64.ZIP. Uncompress that zip file and place it on your workstation C drive. The SqlNexus4.0.0.64 folder will contain the SQL Nexus executable and all the other files needed to analyze the collected data. From there, you can create a shortcut on your desktop or your start menu for sqlnexus.exe.

The DataCollectors subfolder contains the data collector files. There is one ZIP file for each version of SQL Server. In our example, we want to monitor a SQL Server 2012 instance, so we need to uncompress the PerfStatsScript2012.zip file. Once uncompressed, you will see 3 CMD files and their accompanying XML configuration files.

  1. StartSQLDiagTrace2012.cmd & SQLDiagPerfStats_Trace2012.XML
  2. StartSQLDiagDetailed_Trace2012.cmd & SQLDiagPerfStats_Detailed_Trace2012.XML
  3. StartSQLDiagForReplay2012.cmd & SQLDiagReplay2012.XML

StartSQLDiagTrace2012.cmd will likely be the batch file you will run most often. By default, it will collect the most common Performance Monitor counters and run a SQL trace to capture batch level commands. The StartSQLDiagDetailed_Trace2012.cmd batch file will capture the same counters, but it will also capture statement level commands and the Showlan Statistics Profile. The StartSQLDiagForReplay2012.cmd batch file captures the trace events needed to replay it against another server.

Collecting Data
To monitor our server, copy the PerfStatsScript2012 subfolder to the SQL Server. This folder cannot be on a network drive; it must be local to that server. It’s best if you do not place it on the C drive or on a drive that hosts your transaction log (LDF) files. The data collector can quickly capture a lot of data, so you want to avoid filling up the C drive or impacting the write performance of your transaction log files. Open a command prompt to that folder location and run StartSQLDiagTrace2012.cmd. SQLDiag will start up the collectors based on what’s in the XML configuration file. Once you see the green line that says “SQLDIAG Collection started”, then you can try to reproduce your performance problem on the server.

Once you are done, just press Ctrl+C to shut down the data collector. Sometimes it will take several minutes for the collector to fully shutdown, so be patient and do not press Ctrl+C more than once.

You can leave the data collector running for a period of time, but please keep in mind that it can collect a lot of data and the SQL trace can potentially impact your performance. It’s usually best to start it, reproduce your performance issue, and then shut down the data collector.

After you stop the data collector, you will see a sub folder called SQLDiagOutput. Copy that entire folder back to your workstation to begin the analysis of the data.

Analysis of the Data
Run sqlnexus.exe using the shortcut you placed on your desktop. The first step is to specify which SQL Server will be used to perform the analysis. Enter the name of your SQL Server and click Connect.

The tool will automatically create a sqlnexus database on that SQL Server.

Now we need to import the data we collected. In the bottom left hand pane, click Import.

Choose the location of the SQLDiagOutput folder and then click the green Import button. If the collector captured data for more than one instance on the monitored server, then you will get a popup window asking which instance you want to analyze. Choose your instance and click OK.

Depending on how much data was collected, the import process could take a while; maybe even several hours. Be patient and just give it time to complete. While it’s running the Import button will change to a red Stop button. Once it’s complete, it will turn back to a green import button.

Now just close the Data Import window. In the SQL Nexus window, click SQL Perf Main from the upper left hand pane. This is main page where we will drill into the various reports.

Each report name that appears in blue is a hyperlink to that report. Report names that are in black did not have any data to analyze. Most reports have a summary of what they display, but some do not, like the Bottleneck Analysis report. It shows CPU utilization by SQL Server (yellow) and everything else (purple). If CPU utilization is higher for purple, then you could assume the CPU is being used by a process other than SQL Server.

If SQL Server is the higher consumer of CPU (yellow), then look further down on that page to see the wait stats analysis. This is the waits that accumulated while the data collector was running. This will help guide you toward what SQL Server was waiting on while executing the queries. You can click on each wait type to see more details about each wait. If you need to the definition of a wait stat, then look at Books Online.

To go back to the previous page in a report, just click the blue left arrow in the toolbar.

The Perfmon Summary report shows an aggregation of the counters collected from Performance Monitor. They are broken out by CPU, IO, Memory, and Network.

You can click on each blue link to get the graphical view. For example, in the IO report details you can see the Avg Disk Sec/Read and the Avg Disk Sec/Write which will show you how fast your IO subsystem is responding to requests.

The ReadTrace Report is really my favorite group of reports. The main report gives you a graphical view of the overall utilization of resources.

From here, you can drill into the data from various points, such as application name, database id, or login name. The Unique Batches report will show you the top TSQL batches by CPU, reads, writes, and duration. Each graph is inverted, so the one on the bottom is the largest consumer of that resource.

On the Y axis of each graph you will see a number. That number corresponds to the TSQL batch command that is listed further down on the page. For example, Query #1 was the largest consumer of reads. From that table, I can see that was a stored procedure named “SLEV”.

You can click on each batch to get even more detail about it in a graphical view over time.

Below the chart, you can click the + button next to “View Chart Details” to get the detailed info about that TSQL batch. This will display the number executions, CPU, reads, and writes, broken out by interval.

There are many more reports, so be sure to click through each one to view the information they provide.

Customize the Data Collector
By default, the data collectors will capture only a subset of perfmon counters and SQL trace events, and it will do that for ALL instances on the server being monitored. If you need to add or remove certain counters or monitor only a specific SQL Server instance, then you will need to edit the SQLDiagPerfStats_Trace2012.XML configuration file. You can open the file in Notepad, but opening it in SQL Server Management Studio color codes it and make everything easier to read.

The top section of the XML file is for the perfmon stats, which takes up about 75% of the file. The bottom section is for the SQL Server trace events. All of the possible options are listed for every counter; however, the majority are set to disable. Therefore, to add additional counters, just search for the counter name and set the “enabled” value to “true”. For example, if you wanted to add the TSQL:StmtStarting and TSQL:StmtCompleted, then just scroll down to about line 2775 and change the enabled value to “true” and save the XML file. Now when you run the collector it will also collect those two trace events.

If you need to set the name of the server then about line 7 you will see Machine name=”.”. Change the . to the actual server name. If the server being monitored is a Failover Cluster Instance, then you must enter the SQL Server Virtual Server Name.

If you need to set the SQL Server instance name, then about line 2564 you will see Instance name=”*”. Change the * to the instance name. Only type the instance name here, not the ServerNameInstanceName. If your instance to monitor is the default instance, then enter MSSQLSERVER.

Also on the same line, the default authentication to connect to that instance of SQL Server is Windows Authentacation. If you need to use SQL Server Authentication, then set windowsauth to “false” and user to the sql server login name. When you run the StartSQLDiagTrace2012.cmd batch file, it will prompt you to enter the password for that account.

Other Hints and Tricks
Before collecting the data, you need to make sure the Windows account is a member of the local administrators group a member of the sysadmin server role for the server you want to monitor. If you use a SQL Server login to collect the SQL trace events, then it also has to be a member of the sysadmin server role.

Once the data collector is started on the monitored server, don’t log off. That will shut down the collectors and you will have to restart them again.

If you cannot stay logged into the server, then you’ll need to schedule the collection by modifying the StartSQLDiagTrace2012.cmd batch file using the /B and /E parameters for SQLDIAG or register it as a Windows service. Books Online has more information about those options.

When you import data into SQL Nexus, it populates the tables in the sqlnexus database. If you already have data in the sqlnexus database, you’ll get the following error when attempting to import new data.

At this point, you will need to do one of two things: overwrite the existing data or create another sqlnexus database. To overwrite the existing database, click the Import link and then click Options. In the dropdown menu, just select “Drop Current DB Before Importing”. This will drop and recreate the sqlnexus database.

Keep in mind, sqlnexus is just another database on a SQL Server. If you want to keep that data for later analysis, you CAN backup that database before overwriting, or you can create a new database. Before clicking the Import link, click the database dropdown list on the tool bar. From there, you can select “<New Database>”. This just creates another database on your local SQL Server and will be the target of the imported data.

This is the way to go if you need to analyze data from multiple servers, because you can use the dropdown database list to change to a different sqlnexus database and view the reports.

Report Parameters is another feature that people tend to overlook. When viewing a report, such as the IO Metrics, click Parameters button on the toolbar. In the popup window, you can change the start and end timeframe that you want to view. This is extremely helpful if your data spans hours or even days.

Finally, you can export the reports to an Excel file by clicking the Save button on the toolbar.

For example, through your analysis you find the offending query; you can export the batch details report to an Excel file and share that report with the development team or other DBA team members.

Conclusion
As you can see, SQL Nexus can be a very powerful tool to help troubleshoot performance issues within SQL Server. This is why the tool is widely used internally be various SQL Server support teams. The best way for you to get more info is to practice collecting and analyzing data from your servers. However, be sure that you practice on a TEST server and not in your production environment. Soon you’ll be able to quickly diagnose and resolve performance issues in your environment. And one day, it might just save your bacon!

Oct 302012
 

Management Studio is great tool with almost everything you need right at your fingertips.  Notice I said “almost”.  I like to customize the toolbars to meet my needs of doing work every day.  Because being a DBA means you’re busy all day long, and you don’t have time to hunt through all the menus or memorize a keystroke for a special function. 

 
The very first customization I always make is to add the “Hide Results Pane” button to the SQL Editor toolbar.  This option is found under the Window menu and acts a toggle button to either show or hide the results pane for the query editor.
 

 

In SQL Server 2012 Management Studio, from the menu, select Tools, Customize, and then select the Commands tab.   Select the Toolbar radio button, and just to the right of that select SQL Editor from the dropdown menu.  In the Controls section, you will see each command that corresponds to the button on the SQL Editor toolbar.

 

To add a button to this toolbar, click the “Add Command” button.  From the Categories section select Window, then select “Show Results Pane” from the Commands section, and then click OK.  Remember this is a toggle button, so depending on the display state of the results pane, the button with either say “Show Results Pane” or “Hide Results Pane”.  When we’re adding the button in the customization wizard, it always shows up as “Show Results Pane”. 
 
Now you should see the “Show Results Pane” button on the Controls list.

 

 

Make sure that control is selected and click the “Move Down” or “Move Up” button to reposition it on your toolbar.  Once you have the button where you want it, then just click Close.  Your toolbar has now been customized.

 

 

This will save you a few keystrokes and/or mouse clicks when you’re trying to find that one thing that’s buried in the Management Studio menus.  And don’t worry about messing anything up, if you don’t like the layout of a toolbar, you can just click the Reset All button in the Customize wizard to reset the layout back to the default.