Hi, The map-reduce scheduling does take block placement into account and attempts to schedule map tasks accordingly. The system is flexible, and HBase uses it to put maps colocated with the regionservers they read from.
-ryan On Mon, Aug 9, 2010 at 2:32 PM, Renato Marroquín Mogrovejo <[email protected]> wrote: > Hi, thanks for answering, but I did find the answer out. I knew Hadoop did > it, but I needed a reference. > I found it in "Data-intensive Text Processing with MapReduce" in the page > 25. > I quoted here: > > Data/code co-location: . . . "An important optimization here is to prefer > nodes that are on the same rack in the datacenter as the node holding the > relevant data block, since inter-rack bandwidth is significantly less than > intra-rack bandwidth." > > Thanks again. > > > Renato M. > > > 2010/8/9 vaibhav negi <[email protected]> > >> Hi , >> >> I am also a nebie to hadoop. But i think u should read about "rack aware" >> concept. >> It may solve your query. >> >> Vaibhav Negi >> >> >> On Sun, Aug 8, 2010 at 12:10 AM, Renato Marroquín Mogrovejo < >> [email protected]> wrote: >> >> > Hi everyone, >> > Does anybody know if Hadoop can take scheduling decisions using nodes >> > positions >> > within network topology ? Or if there is any work being done on this? >> > Thanks in advanced. >> > >> > >> > Renato M. >> > >> >
