While Spring Batch definitely can do ETL type of processing like what is done in Informatica, the components will not be a one to one mapping. There is a key reason for that. Informatica deals with the entire data set at once (it actually will attempt to load the entire dataset into memory if possible). Where as Spring Batch is item based. Sorting is a data set focused activity so it is not going to be very efficient to do the sort itself in Spring Batch.

Without knowing what your transformations are, I would say that doing things like sorting via SQL in a database will be a better approach than piping it through Spring Batch and having a processor cache and sort the items (which may not even be possible depending on the amount of data).