I think you may schedule your workflow to run continuosly, so it's going to be executed again and again.
Although in this case you need a mechanism to know that data in the source has not been treated yet.
Please, let me know if that helps you in your case.
Thanks for your response. Continuous scheduling will spawn out a new instance every time, which is difficult to maintain when we have to execute the job 24/7. I am looking at a single instance running continuously like a CDC workflow.
PowerCenter works in either of two execution modes, either in the "normal" batch mode or in real-time mode.
Batch mode means you have some "standard" sources (relational, SAP tables/views, files, and the like) which are processed and then finished. This mode essentially reads some source data, processes it, and writes some target data.
Real-Time Mode means you have some so-called real-time source (e.g. a MQ Series message queue, some SAP R/3 IDocs, PowerExchange for CDC, and quite a few more) which delivers data in unknown intervals.
What you are asked to do is definitely a real-time requirement. Using batch-mode sessions and workflows is simply the wrong way to go here, that's all I can say. PowerExchange for CDC would probably be the best option. Or you program a Custom Transformation (programmed in C) to extract those real-time data somehow, then you can even use these data in a batch-mode session (something I've done for a customer in Germany 20 years ago, and that system still runs like hell). Or you program a Java transformation to get those data somehow.
Regarding maintenance I can only advice you to get PowerExchange for CDC, everything else might easily become a nightmare. Just my experience, I don't get paid for these statements.
That's all advice I can offer at this time.