Hi, so you are saying whenever map reduce is executed mapper which takes each regions as its input will run on the same machine where the region is hosted, If wring correct me
Thanks On Tue, May 9, 2017 at 8:32 PM, Kevin O'Dell <ke...@rocana.com> wrote: > Hi Rajeshkumar, > > When you execute a MR job over an HBase table, the reads should be > local(assuming you have good locality now). This is occurs because writes > are sent to the RegionServer that owns the particular Region. Once the > writes occur the block are written to the local DataNode first. Even if > locality gets messed up through RegionServer failures Major Compactions > should restore it. > > On Tue, May 9, 2017 at 8:00 AM, Rajeshkumar J <rajeshkumarit8...@gmail.com > > > wrote: > > > Hi, > > > > > > I have a table with its region max file size is 10 gb. If i run any > > mapreduce over this table whether there will be any data locality. If not > > please explain > > > > Thanks > > > > > > -- > Kevin O'Dell > Field Engineer > 850-496-1298 | ke...@rocana.com > @kevinrodell > <http://www.rocana.com> >
