In server we dont have an option to process the data in multiple nodes as in parallel. In parallel we have an advatage to process the data in pipelines and by partitioning, whereas we dont have any such concept in server jobs.

There are lot of differences in using same stages in server and parallel. For example, in parallel, a sequencial file or any other file can have either an input link or an output ink, but in server it can have both(that too more than 1).