No. There is a key-only filter that you might want to checkout.
We can't skip ahead N rows w/o scanning the intermediates so we know when we've passed out N rows (There is no global index across all regions in the cluster). Skipping to the next region would be the 'natural' thing to do. You can get the complete list up front from HTable. You could make small regions if you wanted to make more discrete jump forwards in your table. St.Ack On Fri, Jun 4, 2010 at 9:22 AM, Geoff Hendrey <[email protected]> wrote: > Hi, > > Is it possible to Scan only the row keys, without bringing back any > columns? > Is it possible to "skip" ahead N rows (rather than retrieving the next N > rows)? > > I am trying to do the most efficient possible scan of a huge table, in > order to make splits for a TableInputFormat. > > -geoff > >
