There are several steps to stopping the task, including:
Stopping the read from the source and marking where the task left off.
Finishing the processing of the current batch of data - applying it to the attrep table (when in batch mode) and then to the target tables - this can take time if the batch is large.
Gracefully logging out from the source and target - the time it takes is dependent upon the number of connections (there can be more than one), the response time of the source & target databases, and the network speed. The source and target endpoint types can also affect the amount of time it takes for the task to stop, for example if there is an intermediate file server before the data is loaded to the target.
You can increase logging for source_capture, target_apply, communication, server, and task_manager if you want to get more details about what the task is doing and which steps are taking longer - then you can work with your source/target/network team to try and address it.
You might also check with your Account Manager on working with Professional Services to tune your tasks so they have the best settings to balance performance while running and a reasonable stop time.
