We have a requirement to scan through the Pega database for over a million cases, extract pertinent information (which would be 5-10 properties) and write this information to an AWS S3 bucket. We want to do this without impacting existing attachments or writes to the current Pega database in our application so as to avoid significant regression testing. Is there a solution / design pattern that is recommended?

We are hosted on cloud but not Pega Cloud.