Can we use different run-time in Copy Activity of Azure Data Factory?

Question

I am using Copy Activity for migrating the Data from On-premises Database to the on-cloud Database. Here I am using Self-Hosted Integration Runtime for both on-premises and on-cloud databases.

The Integration run-time is different for on-premises and on-cloud Databases.

When I execute the pipeline, it shows that both the source and target need to be in the same self-hosted integration runtime.

Is it possible to execute the pipeline having 2 self-hosted integration runtimes?

If it is possible, Please let me know how we can execute the pipeline of having different 2 self-hosted integration runtimes.

Answer 1

Hey,
Ideally whenever an integration run time is used in either source or sink within a copy activity, the necessity is that the integration run time should be able to access both the source and the sink.
So any reason why you want to leverage 2 different IRs?
If you still need it then you would have to create 2 layers:
1st copy data into blob and then copy from blob into destination, similar to how we overcome the onprem Or integration run time limitations in data flows

Answer 2

@Hemanath Goud Kandula , while @Nandan Hegde is correct, I think some additional clarification would be helpful.

You can speficy different integration runtimes in each of the linked services used by a copy activity. However only one actually executes the copy.

A Self-Hosted Integration Runtime (SHIR) is usually used when you have a data store sitting in a network which stops all incoming connections, but allows outgoing connections. Once you place the SHIR inside this network, it makes outgoing connections to Azure to check for jobs.

When a copy activity has 1 SHIR and one Azure IR, the SHIR will take priority because the Azure IR cannot reach into the restricted network.

Now comes the problem of 2 SHIR. One SHIR's outgoing connection would be another IR's incoming connection. Given it is assumed the network denies incoming connections, we run into a problem. Without a direct connection, we cannot move the data directly.

All of this is assuming the data stores are in different networks. If both stores are in the same network, then we can use the same SHIR in both linked services.