mapreduce - In hadoop, I just want to execute my own custom program on each node -
yes, want run custom program on each hadoop node. want deploy no mapper , reducer. distributed computing system doesn't works mapreduce(but uses hdfs internally). how should do?
both mapreduce , tez jobs use yarn (yet resource negotiator) distributed , executed on cluster in so-called containers. can use yarn run own jobs instead. please take @ hadoop architecture overview high-level overview.
Comments
Post a Comment