Find everything you need to get certified on Fabric—skills challenges, live sessions, exam prep, role guidance, and more. Get started
Hi,
This is not so much a Data Story but a template for a dashboard for looking at your own SSIS activity.
I use SSIS extensively with my job and over the years have used various tools to help me manage the packages and triage issues.
These include my own SSIS logs, the out of the box SSMS reports, Jamie Thompson's report pack and a number of SSIS catalog queries.
I had a quite good single query SSIS PowerBI Dashboard, but it was slow on large catalogs and limited in its use.
I started working on new dashboard that provided all of the features from all my past tools and around the time I saw a post from Chris Schmidt (link below) and so decided to use this as a starting point.
My dashboard is still not finished, but after lots of iterations I think it's going to be useful for others who need to monitor and understand what their SSIS packages are doing.
I first posted this a in 2017 and I've just given it an overhaul to support some of the new feature now supported by the August 2018 Desktop & Report server versions. Please give it a try and let me know if you spot any problems or things that could be improved.
I bring in and summarise the data is at a variety of levels
Execution (Execution & Execution Summary) The Package that is Executed First which I've named RootPackageName. This includes the Project & Folder
Package (Executable Package Stats) Package Summary. I group Packages by Types e.g. Master/Dimension/Fact/Stage/PostProcess. You can customised these in the PackageTypes Table. Click Advanced Editor to change it.
Executable (Executable Stats) Items inside the Package e.g. Tasks/Data Flows/Sequence. I also call this level Package Steps
I categorise packages and task into types using a keyword lookup. e.g. Package starting Fact for loading Fact tables.
The Task type names are based on common SSIS naming conventions.
This dashboard also provides details of any package errors and maps them to the Microsoft IS Error Reference Names.
Individual Executions can be viewed as a text based Gantt chart or as a matrix of executions by time slices.
Finally I extracted the Hierarchy from the execution paths to make a Network diagram and Sankey to visualise the structure and levels in an SSIS project.
Whilst developing this dashboard I managed to combine and try out a number PowerBI techniques such as;
Dax Measure to display a Gantt chart (Based on Chris Webb's Star Ratings)
Turn Package Paths into a Network diagram & Sankey Diagram
Scatter Chart to show hour and day matrix
Mini Gantt Timeline with Matrix
Mini Gantt Timeline with Scatter Chart
Dax Tables for Summary data
Vlookup function to use Package names to provide a type. This can do contains matches and use Start and End Indicators.
M Language tables for reference data
Parameterised SQL to Configure the Days to Load
Standard Deviation for tolerances.
Bookmarks & Selections
Report Tooltips
I've tried to keep the custom visuals to a minimum but I have used Sankey/Network Navigator as well as the OKViz Bullet and Sparklines.
The PowerBI is blank with that needs the server parameter setting to your SSIS db server before applying the changes and granting permissions.
The next network and rows can take a while to load wiht a large Catalog db so I have a small version that excludes these.
Templates and sample screen shots are available here https://github.com/stretcharm/SSIS-DB-Dashboard
Release Notes
v1.00
Improved query on Exectutions and Executables
Added Include fields for the Package types to configure the Master & Package Page filtering.
Edit PackageTypesUse Query to configure.
Changed Package filters
Add Date, Start, Execution ID Hierarchy to the Master and Package Pages
Added Preserve Layers Bookmark to help avoid window coming to the front
Remove Optional Job Server Code
Add Icon Ribbon and Help Page.
Minor Fixes
Thanks to the providers of the following pages that I've used to help in the making of this dashboard.
https://blogs.msdn.microsoft.com/sql_pfe_blog/2017/04/18/ssisdb-reporting-with-power-bi/
Chris Schmidt
https://www.excelguru.ca/blog/2015/01/28/creating-a-vlookup-function-in-power-query/
@KenPuls
Star Ratings Quick Measure
http://community.powerbi.com/t5/Quick-Measures-Gallery/Star-Ratings/m-p/166903#M12
@cwebb
https://ssisreportingpack.codeplex.com/
Jamie Thomson
Lots of Dax help and great OK Vis PowerBI Visualisations
http://www.sqlbi.com/
SSIS Catalog DB
https://docs.microsoft.com/en-us/sql/integration-services/service/ssis-catalog
Reza Rad's Article on the SSIS Catalog
http://www.rad.pasfu.com/index.php?/archives/75-SSIS-Catalog-Part-3-Folder-Hierarchy;-Folder,-Projec...
DB Diagram
http://www.rad.pasfu.com/ssis/ssiscatalogpart3/5.png
SSIS Error Codes
https://docs.microsoft.com/en-us/sql/integration-services/integration-services-error-and-message-ref...
Enjoy
Phil
Follow me on twitter https://twitter.com/StretchArm1 for updates
Other gallery entries
PowerBI/SQL Server and Microsoft Business Application Related YouTube Videos
https://community.powerbi.com/t5/Data-Stories-Gallery/PowerBI-SQL-Server-and-Microsoft-Business-Appl...)
SQL Saturday Sessions https://community.powerbi.com/t5/Data-Stories-Gallery/Pass-SQLSaturday-Dashboard/m-p/489529#M2153
Microsoft Bus App Summit & SQL Bits videos
and PowerBI Custom Visual
https://community.powerbi.com/t5/Data-Stories-Gallery/PowerBI-Custom-Visuals/td-p/724185
Most code is also here https://github.com/stretcharm
eyJrIjoiZDM0N2NmMjMtZmE2NS00MDE1LWJkOWQtZmFmZmZkMzI4ZDBkIiwidCI6ImEwYTcyYzIzLTdlMWEtNGYxOC05NDU4LTlhNzUyYTEzMTg1NCJ9
It's showing the Package Durations and Start time/End Times in slices of 15 minutes. So it's a bit like a visual gantt.
The start is reversed so the scatter flows downwards over time.
Below is a sample that shows -6 = 6am to -10 =10am and on the x axis 6.5 = 6:30am and 9.5= 9:30am
Size of the bubbles are durations.
You can click the bubbles to filter the gantt or hover to see details.
This looks like a great way to monitor jobs, thanks for posting. I have downloaded the "SSISDB_Dashboard_v0_81_Full_Blank.pbix" version and can change the msdn connection from localhost to the sql instance that we are looking to monitor. However I cannot find a way to modify the SSISDB connection. The Edit Connection button greys out when I select it.
Any assistance would be greatly appreciated.
Thanks
Change the ServerName Parameter.
This is used for all connections.
Apply should then repoint everything to your server. You should not need to change the SSISdb unless is got a different db name.
Alternativly get the template version from this path as they prompt you to enter the parameters when you open them.
https://github.com/stretcharm/SSIS-DB-Dashboard
If you get errors after changing the refresh each query manully in the query editor to ensure it's happy. Sometimes if needs help to get going. Also note that v81 doesn't like it if you have not executions so make sure you've got something in your SSISDB.
Yes it can be tricky to get the full lineage of SSIS. The SSISDB Catalog doesn't hold this level of data, but it's there in the pacakge xml.
Probably the ideal solution is to use metadata and build packages using BIML. This way your lineage design builds the package so you get documentation and generate the code. BIML is not going to work with complex SSIS requirements though.
Doc Xpress does a decent job of extracting the SSIS guts for documentation. It has lineage functions, but this not as useful as it sounds.
http://help.sentryone.com/docxpress/Index.html?SSISPackage.html
https://www.sentryone.com/products/pragmatic-workbench/doc-xpress/sql-metadata-management
You can also extract code from SSIS using powershell
We are moving to a metadata management of package and source dependancies. Eventually we may add column level lineage, but for now we are doing it on a table/data set level. This will allow us to see which target areas are impacted by any source system issues.
Just noticed that changing servers can cause the Execution to Date Join to be dropped.
Just add it back from ExecutionStartDate to Date
Updated version 0.77 templates (blank docs as I cannot upload templates)
Changed to use Ribbon charts
Fixed Gantt Calculation
Added Some Drillthroughs
Fixed Date join issue
Improved query performance on large SSIS DBs
Small template doesn't include the rows or network diagrams that can be slow.
I've also added a post in the Quick Measures for the Text Gantt Dax
http://community.powerbi.com/t5/Quick-Measures-Gallery/Text-Gantt-Chart/m-p/253466
Updated version 0.78 templates (blank docs as I cannot upload templates).
Use the Server Name Parameter and Apply. Then do a refresh. If you want the Rows manually Refresh Execution Stats Rows as it's not included in the refresh.
Changes.
Added Percentage completion estimates to the Currently Running Tab
Improved query performance on large SSIS DBs
Execution Stats has commented out the time filter as its quicker on my server. If you have a large ssisdb you may want to uncomment.
Improved query for Execution Hierarchy and it's now included in the Refereses
Improved query for Execution Stats Rows. Still not not included in the refresh as it can be slow, but you can refresh manually.
Add a Execution Comparison tab
Add a Execution Over Time tab using scatter play axis
Here are some new screen shots
Hi,
Awesome Report !
How do I input todays date into the parameter so it is dynamic and always looks up to todays date ?
Currently what i am seeing is I have to every morning update the parameter to todays date.
Please advise if I am doing something wrong.
What would be even better if there is any documentation on the solution ?
Regards,
Andrew
Thanks
The report works on a number of days of history to load.
Set the MaxDaysData and the report will always load the latest data.
The example report for the community page works on a single dataset as it's a fixed dataset.
The only documentation is on this thread on in the github page.
https://github.com/stretcharm/SSIS-DB-Dashboard
To use get the template and when it opens populate the parameters to point to your
SSIS db server, Agent Job Server, SSIS db name and the MaxDaysData.
As long as it's SSIS 2012 or later and you have permissions it should just start working.
Start with a small number of days in case there is too much data or the SSIS queries are a bit slow.
If you have more questions I check the forum every once in a while.
Hi
If Possible can please the SQL code for the total package steps and error codes and failed steps.Hariaanem.da2gmail.com is my email. Thank you
@stretcharm wrote:
Updated version 0.78 templates (blank docs as I cannot upload templates).
Use the Server Name Parameter and Apply. Then do a refresh. If you want the Rows manually Refresh Execution Stats Rows as it's not included in the refresh.
Changes.
Added Percentage completion estimates to the Currently Running Tab
Improved query performance on large SSIS DBs
Execution Stats has commented out the time filter as its quicker on my server. If you have a large ssisdb you may want to uncomment.
Improved query for Execution Hierarchy and it's now included in the Refereses
Improved query for Execution Stats Rows. Still not not included in the refresh as it can be slow, but you can refresh manually.
Add a Execution Comparison tab
Add a Execution Over Time tab using scatter play axis
Here are some new screen shots
code is available in the pbix or template files. get them from
https://github.com/stretcharm/SSIS-DB-Dashboard
If you go into the query editor you can see the sql for each area.
This is amazing!
We have an SSIS environment that runs all packages from file system and we do not have an SSISDB. Is this still able to work? I was unable to get it set up because of the missing SSISDB. Packages are all on file system.
Thanks for contributing this!
-Wes
Thanks.
I'm afraid no it only works with SSISDB,
I recommend if your using SQL 2012 or above you try deploying to the catalog as you get all the SSISDB logging for free and it's very good.
In theory you could use file based ssis and create similar logs your self, but it would be lots of work. If they logs are the same as the SSISDB tables you could adapt the powerbi to use files instead of a db.
Is there a reason why your not using the SSIS Catalog?
Phil
Thanks for the info. All of our packages are from a vendor solution so we don't have a whole lot of control over how they are deployed and updated.
OK. You could suggest they get with the times.
If they give you any execution logs you could try building your own dashboard from them.
PowerBI is also very good at loading data from files.
Most of my dashboard is driven from the execution statistic data.
I sent the inquiry today as to whether this is something on their roadmap. I will have to look in to exection logs. The vendor does have their own dashboard for tracking, configuring, and managing the product from an ETL/DW perspective, but they don't have any analytics built in to show trend, day over day, etc. Theorhetically, I could learn their DB well enough to build off of.
Sounds promising. I doubt they have done anything too radically different.
However as long as you can get the package start and ends you've got the basics needed to do the trends.
Hi,
This looks awesome. I wasn't able to run it though as we seem to have too much data (though we only keep 14 days in the catalog). The notes mention a "small" version, but I couldn't figure out where the file for that is in in the download. Is it in a different version?
I'm also getting a duplicate ExecutionID key on the Execution query. I checked the table though and it only exists once. Maybe being duplicated by a join?
Thanks,
Scott
Thanks.
It can sometimes take a bit of getting going on the initials loads. Try ensuring each of the queries work in the query editor.
If you get any errors referring to Formula.Firewall: then try turning off privacy in the options. This can also help performance.
How many projects, ssis executions do you have a day? I have about 180 executions with 8700 steps on a day and can easily load 30 days worth of history. However I tend to use 10 days.
You may want to test the indiviual queries run ok on your environment in Mgmt Studio. Also running index maintainance on the SSIS Catalog db can help. I've tried to optimise based on my data.
There is a small version in the post on the v77. I stopped creating it as I improved the query speed so it was not as bad and it was a pain to keep it up to date.
However it's easy to may a small version, just disable the Load for Execution Path Map. Execution Hierarchy and Execuction Statistics Rows. They are not essential and the queries are slow, especially the rows as it does string maniplation on the messages.. The Network, Sankey and Rows tabs will not be updated.
The Execution Key should be unqiue however check what projects and folders you have. Maybe there is a combination that causes duplciates. If so let me know and I can update the joins.
SELECT ex.[execution_id] AS ExecutionID FROM [catalog].[executions] ex JOIN [catalog].[packages] p ON p.name = ex.package_name JOIN catalog.projects pr ON p.project_id = pr.project_id AND ex.[project_name] = pr.name JOIN catalog.folders f ON pr.folder_id = f.folder_id WHERE ex.[start_time] > DATEADD(DAY, -1* 3, CONVERT(DATE, GETDATE()));
I've also been working on a very cut down version that is for currently running ssis and errors from today which add when I add then next update.
Hi,
We have about 600 executions a day in production, some packages may generate a lot of messages based on their steps, looping or if there were memory errors (not usual but has happened). The event_messages view has about 20 million rows.
I was able to get the queries to start refreshing in test, which has much less activity. That's where I got the duplicate execution key error. It looks like the problem is that I have some situations where packages with the same name exist in multiple places. In one case, I have two folders with virtually the same projects and packages underneath because there were some variations in the packages requiring different code but otherwise the naming still made sense. I think your join from catalog.executions to catalog.packages should be different. It's on package name, which doesn't need to be unique. Perhaps it could be on execution.object_type and execution.object_id. My object_type is always = 20
SELECT ex.execution_id FROM [catalog].[executions] ex JOIN [catalog].[packages] p ON p.package_id = ex.object_id AND ex.object_type = 30 -- Package JOIN catalog.projects pr ON p.project_id = pr.project_id AND ex.[project_name] = pr.name JOIN catalog.folders f ON pr.folder_id = f.folder_id WHERE ex.[start_time] > DATEADD(DAY, -1* 3, CONVERT(DATE, GETDATE())) UNION ALL SELECT ex.execution_id FROM [catalog].[executions] ex JOIN catalog.projects pr ON pr.project_id = ex.object_id AND ex.object_type = 20 -- Project JOIN catalog.folders f ON pr.folder_id = f.folder_id WHERE ex.[start_time] > DATEADD(DAY, -1* 3, CONVERT(DATE, GETDATE()))
Thanks,
Scott