Adam, Thanks for this excellent video. You explained almost every feature available there in data flows. Looking forward a video on Azure SQL DWH. I know it will be great to learn about it from you.
I would say this is the best content I've seen so far!! Thank you so much for making it Adam! Just wondering, is there a Crtl+Z or Crtl+Y command in case we did some changes in the dataflow and restore it to previous version?
Awesome, thanks! Unfortunately not, but you can use versioning in the data factory which will allow you to revert to previous version in case you broke something. Highly recommended. Unfortunately not reverts for specific actions.
@@549srikanth I publish each time I create a significant new step in the pipeline and I use data preview before moving on to the next step. Also, you can , I think, export the code version of the entire pipeline. Presumably you can, then, paste that into a new Pipeline to resurrect your previous version.
Your videos are really great and helped me understand lot of concepts of Azure. Can you please make one using SSIS package and show how to use that within Azure Data Factory
For anyone wondering how to make the year check (or any check) in the second step more robust, you can exchange the following expressions using the 'case' expression as used below which says, if this expression evaluates as true, do this, else do something else. Worth nothing here that in the first expression, there is only a true expression provided while the second expression has both true and false directives. As per the documentation on the 'case' expression: "If the number of inputs are even, the other is defaulted to NULL for last condition." /* Year column expression */ /* If the title contains a year, extract the year, else set to Null */ case(regexMatch(title, '([0-9]{4})'),toInteger(trim(right(title, 6), '()'))) /* title column expression*/ /* If the title contains a year, strip the year from the title, else leave the title alone */ case(regexMatch(title, '([0-9]{4})'),toString(left(title, length(title)-7)), title)
Thanks Paul :) I used as simple example as possible for people who aren't fluent in scala but of course you always need to cover all possible scenarios. Sometimes I like to fail the transformation rather than continue with fallback logic as I expect some values to be present.
@@AdamMarczakYT Of course, I just wanted to see if I could take it a step further to align more closely with what would be needed in a production data engineering scenario and thought others may have the same idea. Thanks for the content! :)
I really like your tutorials. I have been looking for a "table partition switching" tutorial but haven't found any good ones. May be you could do one for us? I am sure it'll be very popular as there aren't any good ones out there and it is an important topic in certifications :-)
Hi Adam, so glad I found your channel. Your videos were a big help for achieving the AZ900 certificate. Now I am studying a lot to uplift my knowledge and get the Azure data engineer certificate. However, I have an important question! Data flows are expensive, sometimes clients don’t want to use this, are there alternatives to achieve the same result in azure data factory? Thank you very much!
@@AdamMarczakYT True! I am currently struggling with csv files that sometimes have extra spaces after the words in the header, this then gives error when doing a copy activity to Azure SQL Database. Do you have any idea to make my flow a bit more flexible so that it can deal with this? It needs some trimming in the header
I thought of doing a SELECT in a dataflow to then change to the correct header titles, but for this I need to know where the spaces will be in the future. So also not flexible.
Thanks for the informative and detailed video adam, 😊👌. Your content is practical. Can you make a video on how load the data from Oracle table to azure data factory? It would be helpful for audiences.
Adam, great video.I m new to Data Flow and I have one doubt, I want to implement File level checks in Data Flow but not able to do it. All tasks are performing data level checks like exist or conditional split. Is it possible to implement File level check like whether file exist or not in storage account?
Adam, FOr using transformation do I need to learn scala. Or just refer the documentation you specified for scala functions and write the transformation?
Hi Adam, just watched two of your videos on Azure Data Factory, nice work. Any chance you can do one on ADF using REST API as a data source with a JSON output, then store in a SQL Server sink?
Very Nice video .... I have one question ... How cluster will decide the number of worker node and driver node it will require in a cluster for running the data flow?
By default it's 4 core single worker node cluster but you can change that in Dataflow action in the settings tab. There you choose from 4 worker cores and 4 driver cores to 256 worker cores and 16 driver cores. Although I'm not entirely how many nodes does it spawn. As far as I understand for databricks it's always one driver node. Check here for some extra info docs.microsoft.com/en-us/azure/data-factory/concepts-data-flow-overview#azure-integration-runtime-data-flow-properties
Hello Adam , i follow these steps but i have a problem : i didn't find the source columns when i go to derived column component to write expression based on existing column. in your video , total columns in source component show = 3 , for me =0 ? i changed the source from csv to sql table and i didn't found the solution.
Some SAP products are supported, always check official documentation docs.microsoft.com/en-us/azure/data-factory/connector-overview?WT.mc_id=AZ-MVP-5003556
Hello Adam. Your video is impressive, as always, but I'm concerned about the source dataset. Question: Does the DataFlow activity only work if the datsource are connected to Azure SQL? I tried using a previous dataset connected to the local server, but this dataset does not appear on the Source settings / Source options / Source dataset in DataFlow activity option. I tried with New option and it is only enabled to select the AZURE dataset. All options in the database are disabled. So I couldn't create a data set for SQL Server neither.
Hey mapping data flows currently support 6 data services for both source and sink. docs.microsoft.com/en-us/azure/data-factory/data-flow-source#supported-source-connectors-in-mapping-data-flow I'd check if you can trick data flows by using Azure SQL connector to connect to on premise SQL server, but I never personally tried.
Maybe, but low-code solutions allow for data scientists and people with less technical knowledge of programming languages to perform data transformations. Code isn't always the best way, in fact for some apps it's not a good thing. :)
A quick question, Azure dataset seems only support already structured data, like CSV or JSON, what if my datasource is an unstructured text file that must be transformed into csv before being used? Is there a way to do this transformation (possibly python code) in data factory?
Hi Mr adam how are you? been trying to use the add function to add two columns of numeric value but the result is wrong E.G ADD(COLUMN_A, COLUMN_B) RESULT =COLUMN_AB instead of adding the values. lets say column_a have value of 334 and coumn_b have value of 4 result is giving 3344 instead of 338. please can you help. Nice video BTW. thanks
Adam, I have been watching many of your videos. As someone new to Azure, i find your videos immensely valuable. Keep up your great work, really appreciate!
Adam, is there a way to preserve the filename and just have it change the extension? For instance, I'm adding a column with datetime, but at the end I would like it to have the same file name, just parquet. Is there a way to do that?
@@AdamMarczakYT Sorry if it was a dumb question, I'm still new to ADF. Ignore if it's too inane but is fileanem in the @pipeline parameter? I found one online but couldn't get it to parse.
I have a question, please help. I am not able to understand why DataFlows need to have their own data sets. Why not use the pipeline datasets. This will help me a lot. Thanks in advance.
Hi Adam, is it possible to create these pipelines as code as well? Or somehow create them from my actual Azure pipeline? It would be sheerly insane (but it is a Microsoft product) to require and maintain two pipeline one that’s yiur Azure pipeline for CI and CD and one for the ADF. I really would want the Azure pipeline to be able to fill/create the ADF pipeline. But I haven’t found anything yet.
Hello Adam, thanks a bunch for this excellent video. The tutorial was very thorough and anyone new can easily follow. I do have a question though. I am trying to replicate an SQL query into the Data Flow, however, I have had no luck so far. The query is as follows: Select ZipCode, State From table Where State in ('AZ', 'AL', 'AK', 'AR', 'CO', 'CA', 'CT'...... LIST OF 50 STATES); I tried using Filter, Conditional Split and Exists transforms, but could not achieve the desired result. Being new to the Cloud Platform, I am having a bit of trouble. Might I request you please cover topics like Data Subsetting/Filtering (WHERE and IN Clauses etc.) in your tutorials. Appreciate your time and help in putting together these practical implementations.
How do you delete from target based on data from the Source? I'm really struggling to understand if i have a column with a value that I want to delete in the target table. Everything seems to be geared up to altering source data coming in
hi , does Azure Data factory can be used to Replace IBM DataStage Mappings transformation. as ibm datastage is a etl tool and azure data factory is a managed data integration service on cloud. does azure data factory supports only blob storarage , azure cosmos db (sql api) , azure data lake storage , azure sql data warehouse azure sql database only ? Apart from these , does Azure Data factory connects to SAP HANA , SAP bw , oracle . are there any connectors being used to pull data from other sources like SAP hana ,oracle etc
Hey, in general ADF has 80+ connectors, including SAP and Oracle. You use those to copy data from those sources to blob storage and then trigger mapping data flow pipeline to get data from blob storage (or data lake), transform it and output it back to blob (or one of supported output systems). Where ADF copies it to designated place.
Adam, your tutorial are very simple to follow and provides lot of insight. However, I could not set up my data flow in the way I wanted. What I want to do is -> Get a list of integer from Table 1 (on-prem SQL server) -> Use this list of integer to query a cosmos structured stream -> Then I want to perform some transformation with the data returned from this cosmos stream -> sink transformed data back to Table 2 (on-prem SQL server). However, I do not see my source datasets on the data flow drop down. Any help would be appreciated.
As useful another Awesome video Adam !!!. Excellent. It was to the POINT !!!. Keep up the good work which you have been doing for plenty of users like me. Eagerly waiting for more similar videos like this from you !!!. Can you please have some videos for Azure Search ...
Thank you so very much :) Azure Search is on the list but there is so many news coming from Ignite that I might need to change the order. Let's see all the news :).
Here is the list of supported data sources for MDF docs.microsoft.com/en-us/azure/data-factory/data-flow-source?WT.mc_id=AZ-MVP-5003556 . Just copy data from REST API to Blob and then start MDF pipeline using that blob path as a parameter.
I call foul. There is a space between movie title "Toy Story" and the year "(1995)". The formula should return "Toy Story " with a space on the end not "Toy Story". But reviewing the output, the space is trimmed off regardless of these 2 expressions: left(title,length(title)-6) left(title,length(title)-7) Both yield the exact output of "Toy Story" with no space. This is not ok.
Your video content is awesome!!! Your video is very useful to understand Azure concept specially for me who just started Azure journey. I would like to have one video where we can see how to deploy code from Dev to QA to Prod. How to handle connection string, parameter etc during deployment. thanks again for wonderful video content.
ADF CI/CD is definitely on the list. It's a bit complex topic to get it right so it might take time to prepare proper content around this. Thanks for watching and suggesting ;)
Hi Adam, few doubts. Please help me understand. 1. 10:04, After running the dataflow 1st time, there are 9125 rows got populated. Well, there is no output sink or output dataset associated with it dataflow yet, then where exactly those ingested rows are getting saved/populated? 2.15:04, after re-calculating "title" (by removing the year part), how come the previous original column (title) got disappeared? The modified title column should appear in addition to the previous original column (title) right?
It's the same as blob storage, just create linked service and select Azure Table Storage and create dataset for it. Not that this is not supported for Mapping Data Flows.
Hi, I have an issue and I am wandering if you could help me. I have created similar data flow. When I run a pipeline with this data flow inside I can see that new file was created in my Data lake. Unfortunately, this file is always empty, but when I click preview data within data flow (in a sink tool) I can see data. Empty means it has column names, but it doesn't have any data. This file is CSV
Which file are you checking? Mapping Data Flow creates many files in the output to follow partitioned model which is HDFS compatible. Typically there is an empty file and a folder which inside contains partitioned data.
Hi Adam, please add some more contents about new features of dataflow, it's your channel only where I see azure add, no one teaches better than you do as I have compared with many channels.
I have an issue with the column 'title' not being found in the derived column despite being able to see all the column in the source beforehand.. Very confused!
Can I do the same process with SQL? how does the partition work in this case? Separate into SQL DataBase or just creates one single table with the results?
You mean using data flows with on premise SQL? Well data flow can't run on self-hosted integration runtime so your best bet is to copy data to Azure blob, transform it and then put it back to on prem sql. Remember underneath this, it's databricks so data flows are not running on SQL.
Great video. Question: Under "New Datasets", is there a capability to drop data into Snowflake? I see S3, Redshift, etc. I appreciate the video and feedback!
Hi Adam, Thanks for making this videos, very clear and concise. I have a question (sorry not related to this video) regarding Conditional split - Can the output stream activities, run in parallel ?
There are some properties available in the trigger which can be used. Example @trigger().startTime docs.microsoft.com/en-us/azure/data-factory/concepts-pipeline-execution-triggers?WT.mc_id=AZ-MVP-5003556#trigger-type-comparison
I love you, Adam! I have been struggling with using expression builder in Data Flow. I can't seem to figure out how to write the code. This video just made it look less complex. I'll be devoting more time to it.
Check if you selected 'first row as headers' checkbox in your dataset. Also there is a button preview data there which you can use to check if you properly set up your dataset. If all will fail, I often advise my colleagues to just delete everything and start over, while sounds a bit weird it really sometimes is faster than finding the issue and provides nice learning curve, Good luck!
Great video! Most videos seem to focus mostly on the evertisement material straight from Azure. At best they show you the very dumb step of copying data from a file to DB. This is the first video I saw where you actually show how you can do something useful with the data and close to real life scenario. Thank you.
Good catch I forgot to add this to description. You can find the file here. github.com/MarczakIO/Azure4Everyone/tree/master/AzureDataFactory-MappingDataFlow-Intro
![Power BI Dataflows Tutorial and Best Practices [Full Course] 📊](http://i.ytimg.com/vi/GuU6_cqu8FE/mqdefault.jpg)
This is quality stuff. Good for a quick upskill especially when prepping for an interview.
very good explanation Adam. keep it up.
Thanks, will do!
@@AdamMarczakYT Adam do we have trail version of Azure for Learning purpose?
Hi Adam, Thank for helping us in learning new technologies. You are awesome 👌🏻👌🏻👌🏻👏👏.
My pleasure!
Very good explaining the Data Flow. Thanks Mr.Adam.
very well done on explaining principles of mapping data flows!!!
Thanks a lot!
Thanks buddy ...Great work
My pleasure
Adam, Thanks for this excellent video. You explained almost every feature available there in data flows. Looking forward a video on Azure SQL DWH. I know it will be great to learn about it from you.
Glad it was helpful! I'm just waiting for new UI to come to public preview then the video will be done :)
Adam you have an ability at explaining complex things, this tutorial made my day, thanks
Glad it helped! Thanks!
I would say this is the best content I've seen so far!! Thank you so much for making it Adam!
Just wondering, is there a Crtl+Z or Crtl+Y command in case we did some changes in the dataflow and restore it to previous version?
Awesome, thanks! Unfortunately not, but you can use versioning in the data factory which will allow you to revert to previous version in case you broke something. Highly recommended. Unfortunately not reverts for specific actions.
@@AdamMarczakYT Excellent!! Thank you so much for your reply!
@@549srikanth I publish each time I create a significant new step in the pipeline and I use data preview before moving on to the next step. Also, you can , I think, export the code version of the entire pipeline. Presumably you can, then, paste that into a new Pipeline to resurrect your previous version.
Your videos are really great and helped me understand lot of concepts of Azure. Can you please make one using SSIS package and show how to use that within Azure Data Factory
For anyone wondering how to make the year check (or any check) in the second step more robust, you can exchange the following expressions using the 'case' expression as used below which says, if this expression evaluates as true, do this, else do something else.
Worth nothing here that in the first expression, there is only a true expression provided while the second expression has both true and false directives. As per the documentation on the 'case' expression: "If the number of inputs are even, the other is defaulted to NULL for last condition."
/* Year column expression */
/* If the title contains a year, extract the year, else set to Null */
case(regexMatch(title, '([0-9]{4})'),toInteger(trim(right(title, 6), '()')))
/* title column expression*/
/* If the title contains a year, strip the year from the title, else leave the title alone */
case(regexMatch(title, '([0-9]{4})'),toString(left(title, length(title)-7)), title)
Thanks Paul :) I used as simple example as possible for people who aren't fluent in scala but of course you always need to cover all possible scenarios. Sometimes I like to fail the transformation rather than continue with fallback logic as I expect some values to be present.
@@AdamMarczakYT Of course, I just wanted to see if I could take it a step further to align more closely with what would be needed in a production data engineering scenario and thought others may have the same idea. Thanks for the content! :)
Thanks, I bet people will appreciate this :)
Just discovered the channel. Your material is hight quality. It's excellent work. I will go watch more. Thank you Adam !
Thank you. This means much :)
Hello Adam, pls let me know how to connect to dynamic crm .. Pls send detail to pradysg@gmail.com
I really like your tutorials. I have been looking for a "table partition switching" tutorial but haven't found any good ones. May be you could do one for us? I am sure it'll be very popular as there aren't any good ones out there and it is an important topic in certifications :-)
Outstanding.
Thanks Mike ;)
Love these videos so easy to understand, do you have a video on new XML connector
Great, thanks! Not yet, maybe in near future :)
Hi Adam, so glad I found your channel. Your videos were a big help for achieving the AZ900 certificate. Now I am studying a lot to uplift my knowledge and get the Azure data engineer certificate. However, I have an important question! Data flows are expensive, sometimes clients don’t want to use this, are there alternatives to achieve the same result in azure data factory? Thank you very much!
Well you can't have the cookie and eat the cookie :) In my opinion it's not that expensive compared to other available tools.
@@AdamMarczakYT True! I am currently struggling with csv files that sometimes have extra spaces after the words in the header, this then gives error when doing a copy activity to Azure SQL Database. Do you have any idea to make my flow a bit more flexible so that it can deal with this? It needs some trimming in the header
I thought of doing a SELECT in a dataflow to then change to the correct header titles, but for this I need to know where the spaces will be in the future. So also not flexible.
Awesome again.
Glad you think so!
very good explanation..keep doig
Thanks. Will do.
very nice
Thanks for the informative and detailed video adam, 😊👌. Your content is practical. Can you make a video on how load the data from Oracle table to azure data factory? It would be helpful for audiences.
Thanks! That's the plan :)
Great video thank you
Thanks again! :)
Adam, great video.I m new to Data Flow and I have one doubt, I want to implement File level checks in Data Flow but not able to do it. All tasks are performing data level checks like exist or conditional split. Is it possible to implement File level check like whether file exist or not in storage account?
Adam, FOr using transformation do I need to learn scala. Or just refer the documentation you specified for scala functions and write the transformation?
Documentation should be enough. MDF is targeting simple transformations so in most cases documentation alone will suffice.
can you pyspark or sql in Expression functions ?
are only scale
Hi Adam, just watched two of your videos on Azure Data Factory, nice work. Any chance you can do one on ADF using REST API as a data source with a JSON output, then store in a SQL Server sink?
Great suggestion! I'll add it to the list of potential topics. :) thanks for watching ;)
Can you how to add the aggregation column to the same output?
You need to use join step and join it with the main output.
Very Nice video .... I have one question ... How cluster will decide the number of worker node and driver node it will require in a cluster for running the data flow?
By default it's 4 core single worker node cluster but you can change that in Dataflow action in the settings tab. There you choose from 4 worker cores and 4 driver cores to 256 worker cores and 16 driver cores. Although I'm not entirely how many nodes does it spawn. As far as I understand for databricks it's always one driver node. Check here for some extra info docs.microsoft.com/en-us/azure/data-factory/concepts-data-flow-overview#azure-integration-runtime-data-flow-properties
@@AdamMarczakYT : stay safe ...stay healthy
Thanks, and same to you! Stay safe my friend!
Hello Adam , i follow these steps but i have a problem : i didn't find the source columns when i go to derived column component to write expression based on existing column. in your video , total columns in source component show = 3 , for me =0 ? i changed the source from csv to sql table and i didn't found the solution.
does DFT support SAP as a data source?
Some SAP products are supported, always check official documentation docs.microsoft.com/en-us/azure/data-factory/connector-overview?WT.mc_id=AZ-MVP-5003556
@@AdamMarczakYT thank you. and is there any way that one can access files from linux os based system as we cant install self hosted IR on linux?
Hello Adam. Your video is impressive, as always, but I'm concerned about the source dataset. Question: Does the DataFlow activity only work if the datsource are connected to Azure SQL?
I tried using a previous dataset connected to the local server, but this dataset does not appear on the
Source settings / Source options / Source dataset in
DataFlow activity option. I tried with New option and it is only enabled to select the AZURE dataset. All options in the database are disabled. So I couldn't create a data set for SQL Server neither.
Hey mapping data flows currently support 6 data services for both source and sink.
docs.microsoft.com/en-us/azure/data-factory/data-flow-source#supported-source-connectors-in-mapping-data-flow
I'd check if you can trick data flows by using Azure SQL connector to connect to on premise SQL server, but I never personally tried.
Wouldn't it be simpler to do all of this using code.
Maybe, but low-code solutions allow for data scientists and people with less technical knowledge of programming languages to perform data transformations. Code isn't always the best way, in fact for some apps it's not a good thing. :)
I owe you my paycheck tbh 😅🤣
A quick question, Azure dataset seems only support already structured data, like CSV or JSON, what if my datasource is an unstructured text file that must be transformed into csv before being used? Is there a way to do this transformation (possibly python code) in data factory?
Hey, you can call azure databricks which can transform any file using Python/Scala/R etc. But data factory itself can't do it.
@@AdamMarczakYT Got it. Thanks a lot! It looks like I have to learn Spark :-)
Hi Mr adam how are you? been trying to use the add function to add two columns of numeric value but the result is wrong
E.G ADD(COLUMN_A, COLUMN_B) RESULT =COLUMN_AB instead of adding the values. lets say column_a have value of 334 and coumn_b have value of 4 result is giving 3344 instead of 338. please can you help. Nice video BTW. thanks
Check out concat function docs.microsoft.com/en-us/azure/data-factory/data-flow-expression-functions#concat
Adam, I have been watching many of your videos. As someone new to Azure, i find your videos immensely valuable. Keep up your great work, really appreciate!
Awesome, thank you!
instead of scala functions is there a way we can use the pyspark functions for debug, BTW these are great videos thankyou
Unfortunately not at this time. if you need more complex constructs or different languages you need to use Databricks or HDInsight :)
Ciekawe filmy, czy mozna znalezc na Twoim kanale film: ADF - Azure Batch Account - Python ? Z pozdrowieniami,
Dzięki, niestety nie mam nic pod batch account z ADF. Rzadko wykorzystywany przypadek u mnie.
It must be very challenging to do all this thing in English for you I imagine, Adam! Congratulations for pushing through despite the difficulty. 🙂
Amazing Video, we want other parts !
Adam, is there a way to preserve the filename and just have it change the extension? For instance, I'm adding a column with datetime, but at the end I would like it to have the same file name, just parquet. Is there a way to do that?
Use expressions :) That's what they are for.
@@AdamMarczakYT Sorry if it was a dumb question, I'm still new to ADF. Ignore if it's too inane but is fileanem in the @pipeline parameter? I found one online but couldn't get it to parse.
Nice video.
Just curious. Can you explain toInteger(trim(right(title,6),'()')) in detail please. Like how this command executes?
I have a question, please help. I am not able to understand why DataFlows need to have their own data sets. Why not use the pipeline datasets. This will help me a lot. Thanks in advance.
It can use pipeline datasets, but not all types/source systems are supported.
@@AdamMarczakYT Thanks Adam, and your videos are very nice. Keep it up.
Outstanding !You just made Azure easy to learn. Thank you.
Awesome, thank you!
ADF is but just one part of about 100 significant tools and actions in Azure. :-(
Hi Adam, is it possible to create these pipelines as code as well? Or somehow create them from my actual Azure pipeline? It would be sheerly insane (but it is a Microsoft product) to require and maintain two pipeline one that’s yiur Azure pipeline for CI and CD and one for the ADF. I really would want the Azure pipeline to be able to fill/create the ADF pipeline. But I haven’t found anything yet.
Hello Adam, thanks a bunch for this excellent video. The tutorial was very thorough and anyone new can easily follow. I do have a question though. I am trying to replicate an SQL query into the Data Flow, however, I have had no luck so far.
The query is as follows:
Select ZipCode, State
From table
Where State in ('AZ', 'AL', 'AK', 'AR', 'CO', 'CA', 'CT'...... LIST OF 50 STATES);
I tried using Filter, Conditional Split and Exists transforms, but could not achieve the desired result. Being new to the Cloud Platform, I am having a bit of trouble.
Might I request you please cover topics like Data Subsetting/Filtering (WHERE and IN Clauses etc.) in your tutorials.
Appreciate your time and help in putting together these practical implementations.
-1979 and ,12
This is why complex logic is needed. Nice tutorial :)
an error message e.g. handshake_failure when the data flow source retrieve data from API, can anyone help? thanks.
How do you delete from target based on data from the Source? I'm really struggling to understand if i have a column with a value that I want to delete in the target table. Everything seems to be geared up to altering source data coming in
Video is excellent. I want to know the problem statement which Data flow is solving?
Would you plan to make video for introduction of each transforamtion components? Thanks
Does any of these option changed now? Because I am not able to see any data debug option to be enabled, and directly preview data in dataset itself.
Will it work with pipe (“|”) separated value file instead of csv?
Thank you Adam Dzienkuje, this is a great tutorial.
Great video! Thanks Adam!
My pleasure!
Very useful. Thank you so much.
Glad it was helpful!
hi , does Azure Data factory can be used to Replace IBM DataStage Mappings transformation. as ibm datastage is a etl tool and azure data factory is a managed data integration service on cloud. does azure data factory supports only blob storarage , azure cosmos db (sql api) , azure data lake storage , azure sql data warehouse azure sql database only ? Apart from these , does Azure Data factory connects to SAP HANA , SAP bw , oracle . are there any connectors being used to pull data from other sources like SAP hana ,oracle etc
Hey, in general ADF has 80+ connectors, including SAP and Oracle. You use those to copy data from those sources to blob storage and then trigger mapping data flow pipeline to get data from blob storage (or data lake), transform it and output it back to blob (or one of supported output systems). Where ADF copies it to designated place.
Adam, your tutorial are very simple to follow and provides lot of insight. However, I could not set up my data flow in the way I wanted.
What I want to do is -> Get a list of integer from Table 1 (on-prem SQL server) -> Use this list of integer to query a cosmos structured stream -> Then I want to perform some transformation with the data returned from this cosmos stream -> sink transformed data back to Table 2 (on-prem SQL server). However, I do not see my source datasets on the data flow drop down. Any help would be appreciated.
Make sure to check if you use supported data source types docs.microsoft.com/en-us/azure/data-factory/data-flow-source?WT.mc_id=AZ-MVP-5003556
As useful another Awesome video Adam !!!. Excellent. It was to the POINT !!!. Keep up the good work which you have been doing for plenty of users like me. Eagerly waiting for more similar videos like this from you !!!.
Can you please have some videos for Azure Search ...
Thank you so very much :) Azure Search is on the list but there is so many news coming from Ignite that I might need to change the order. Let's see all the news :).
Great Video. Can you use data from a REST Api as a source for a Mapping Data Flow or does the source have to be a dataset on Azure?
Here is the list of supported data sources for MDF docs.microsoft.com/en-us/azure/data-factory/data-flow-source?WT.mc_id=AZ-MVP-5003556 . Just copy data from REST API to Blob and then start MDF pipeline using that blob path as a parameter.
I call foul. There is a space between movie title "Toy Story" and the year "(1995)". The formula should return "Toy Story " with a space on the end not "Toy Story". But reviewing the output, the space is trimmed off regardless of these 2 expressions:
left(title,length(title)-6)
left(title,length(title)-7)
Both yield the exact output of "Toy Story" with no space. This is not ok.
You can always use ltrim function too! :) The expressions and data used in the demo is from Microsoft's example.
Wow..lucid explanation..
Glad you think so!
Your video content is awesome!!! Your video is very useful to understand Azure concept specially for me who just started Azure journey.
I would like to have one video where we can see how to deploy code from Dev to QA to Prod. How to handle connection string, parameter etc during deployment.
thanks again for wonderful video content.
ADF CI/CD is definitely on the list. It's a bit complex topic to get it right so it might take time to prepare proper content around this. Thanks for watching and suggesting ;)
Thank you, very helpful tutorials
Hi Adam, few doubts. Please help me understand.
1. 10:04, After running the dataflow 1st time, there are 9125 rows got populated. Well, there is no output sink or output dataset associated with it dataflow yet, then where exactly those ingested rows are getting saved/populated?
2.15:04, after re-calculating "title" (by removing the year part), how come the previous original column (title) got disappeared? The modified title column should appear in addition to the previous original column (title) right?
hey 1. it's amount of rows loaded. 2. if you create new column with the same name it will replace old one. In this case we replaced title column.
can you please explain who to connect source dataset from azure data lake storage gen 2 tables in data flows of Azure data factory?
It's the same as blob storage, just create linked service and select Azure Table Storage and create dataset for it. Not that this is not supported for Mapping Data Flows.
Your way of explaining is outstanding, after watching it feel like Azure is very easy to learn. kindly keep sharing good videos Thank You..
Thanks a ton :)
Wow,I like your video, I did it today. and I had good result. thanks for your good explanation.
Great job! Thanks!
Hi, I have an issue and I am wandering if you could help me. I have created similar data flow. When I run a pipeline with this data flow inside I can see that new file was created in my Data lake. Unfortunately, this file is always empty, but when I click preview data within data flow (in a sink tool) I can see data. Empty means it has column names, but it doesn't have any data. This file is CSV
Which file are you checking? Mapping Data Flow creates many files in the output to follow partitioned model which is HDFS compatible. Typically there is an empty file and a folder which inside contains partitioned data.
Hi Adam, please add some more contents about new features of dataflow, it's your channel only where I see azure add, no one teaches better than you do as I have compared with many channels.
Thank you! What kind of features would you think would be interesting to see?
Very nice tutorial 👍
Thank you! Cheers!
Amazing videos.
Glad you think so! :)
I have an issue with the column 'title' not being found in the derived column despite being able to see all the column in the source beforehand.. Very confused!
When attempting to aggregate - no columns are found. Again despite seeing them in the source.
I've rebuilt the whole thing and still face the same issue. Google yields no results either.. Does anyone know what i'm doing wrong?
Just tried again and it works! The only difference this time round was that I didn't enable data flow debug. No idea why it worked this time.
Also 'Actions' no longer exists under pipeline - Is there a new way to view the details pane? I can't seem to find one.
These actions can be now found if you hover over 'Name'!
Thank you so much Adam! this was very clear and great video and a big help for my interview and knowledge.
Very welcome! Thanks for stopping by :)
nice & detailed video.
Thank you!
Can I do the same process with SQL? how does the partition work in this case? Separate into SQL DataBase or just creates one single table with the results?
You mean using data flows with on premise SQL? Well data flow can't run on self-hosted integration runtime so your best bet is to copy data to Azure blob, transform it and then put it back to on prem sql. Remember underneath this, it's databricks so data flows are not running on SQL.
excellent explanation with simple scenario. Thank you.
Glad it was helpful!
best tutorial ever... 💪🏻💪🏻💪🏻
Great video.
Question: Under "New Datasets", is there a capability to drop data into Snowflake? I see S3, Redshift, etc.
I appreciate the video and feedback!
These videos are great. Helping me so much! Thanks Adam
Glad you like them!
I just find your videos while searching for ADF tutorials in youtube. The materials are fantastic and really helping me to learn. Thank you so much!!
Happy to help! :)
Hi Adam, Thanks for making this videos, very clear and concise. I have a question (sorry not related to this video) regarding Conditional split - Can the output stream activities, run in parallel ?
They typically run in parallel as it's Apache Spark behind the scenes.
@@AdamMarczakYT Thank you !
Lovely bro!!
Thanks 🔥
if I chage the value of the date ,how can I do it with everyday for tigger time. I mean pass the datetime by tigger time as blob path
There are some properties available in the trigger which can be used.
Example @trigger().startTime
docs.microsoft.com/en-us/azure/data-factory/concepts-pipeline-execution-triggers?WT.mc_id=AZ-MVP-5003556#trigger-type-comparison
Thanks adam
My pleasure!
Your videos are very informative and practical oriented. Keep doing .
Thank you, I will!
Brilliant way of explanation
Subscribed to your channel
Thank you, appreciated 🙏
This was explained very well. thank you.
You're very welcome!
I love you, Adam!
I have been struggling with using expression builder in Data Flow. I can't seem to figure out how to write the code. This video just made it look less complex. I'll be devoting more time to it.
Hi Adam,
can you suggest me how to perform custom transformations on our data?
If Mapping Data Flow doesn't suffice then maybe try Azure Databricks. Check my tutorial and decide yourself. I use it in pretty much any project. :)
I got stuck at Derived Column, since not able to get columns from source (Movies input)
Check if you selected 'first row as headers' checkbox in your dataset. Also there is a button preview data there which you can use to check if you properly set up your dataset. If all will fail, I often advise my colleagues to just delete everything and start over, while sounds a bit weird it really sometimes is faster than finding the issue and provides nice learning curve, Good luck!
I have the same issue - did you resolve it?
Thanks
You're welcome!
Great video! Most videos seem to focus mostly on the evertisement material straight from Azure. At best they show you the very dumb step of copying data from a file to DB.
This is the first video I saw where you actually show how you can do something useful with the data and close to real life scenario.
Thank you.
could you please share me the csv file that ur using fo this demo..
Good catch I forgot to add this to description. You can find the file here.
github.com/MarczakIO/Azure4Everyone/tree/master/AzureDataFactory-MappingDataFlow-Intro
Thanks for such good video
very very detailed work flow , i tried this and able to understand Data flow process so easily . Thank you for the wonderful session.
Hi Adam, what is the language for dataflow ? how to find more resouce for it ?
It's scala :)
best video on azure I have ever seen❤❤
Nice one Adam. Cool one. Keep doing fabulous videos always fella.
Many THanks.
Great! You are the best Adam.
How do you solve the parallel execution of your pipeline when triggered by events to avoid duplicates?
You need to do this as part of your flow design. Unfortunately some things can't be solved by tools. Thanks for watching! :)
@@AdamMarczakYT Ok maybe handle it by Run ID I guess :)
Good explanation there.