paging large amounts of data

Hi, its not entirely clear to me how paging of large amounts of data are going to work. essentially the way things seem to me is that I need to stick in an array with all the id's of the entire dataset, even if in the end I am only interested in a small number of all the id's and even if I am able to specifically extract these ids using LIMIT (or some equivalent syntax) from the database. Is there a way to just pass in a count of the total number of items in the dataset instead? It does not seem sensible to fake this huge array either. regards, Lukas