Search code examples
azureazure-data-factoryazure-synapse

Best approach to trigger Synapse from ADF

I would like to hear from you if my approach is the best out there, or if there are some out of the box solution or better solution when it comes to convenience and security. I have an Azure Data Factory serving as the ingestion tool and there is Synapse Analytics as the transformation (my solution is mostly in Synapse but you cannot share Self Hosted Integration Runtime linked service among synapse workspaces so it would require more amount of VM with SHIR linked services which has been denied as its cost).

Currently the only way I see is WebActivity within ADF pipeline and REST API call to trigger Synapse pipeline. I guess it's pretty decent solution as I've come across that before, but maybe there is something better I can think of? Any recommendation and idea would be highly appreciated

Thank you in advance

Solution

  • Currently, there is no specific activity in ADF to trigger the synapse pipeline. As mentioned, you can either use a REST API to trigger the synapse pipeline or use storage event trigger.

    • The Storage event trigger approach requires a temporary storage location where no files apart from ADF should be uploaded to this. Create storage event trigger in the synapse pipeline to this location. In the ADF pipeline, after all the activities, use a copy activity and copy a sample file to this location. Every time, you want to trigger the synapse trigger, you can run the ADF pipeline, and it will copy the same sample file to this storage location which will trigger the synapse pipeline. The sample file can be of any type as it will be used only for the triggering. In this approach, you can avoid the possibility of hardcoding the secrets and Bearer tokens.
    • The REST API approach requires hardcoding of the secrets and Bearer tokens, but you can use the option of securing the input and output of the web activity. Additionally, in this approach, you can pass any parameters if you want from the web activity body. And also, if you return any values from the synapse pipeline, you can capture the output values after the synapse pipeline run. Go through this SO answer to know more about it.