On Fri, Oct 29, 2010 at 7:44 AM, Sean Bigdatafun <[email protected]> wrote: > I think the low-latency guarantee (or at least in some degree) requirement > prevents any heavy M/R job in the same cell, and here is the reason: > ---- If a heavy M/R task gets started to run on a machine, it may peg > the CPU, evict memory and so on, which basically makes the access to data > belonging to that RS much higher latency than normal. > > Any comment? > >
Yes. A heavy-mapreduce job will effect any other service running on same hardware whether it hbase or anything else. This is not to say you cannot run mapreduce at all on your low-latency cluster. For example, scans are relatively low impact and can be configured to be even less so if asked to not populate block cache as they run . Also the quotidian cluster load is often sinusoidal. Heavier mapreduce jobs can be run at downtimes. St.Ack
