Several questions to ask about hadoop:

 - when I install hadoop under /usr/local/hadoop/hadoop for master. Can
other slaves be installed under different directories? Or every machine
should have the same installation directory?

 - After I successfully installed hadoop, how do I run real jobs on it?
Should every job be written in java, as the wordcount example?

 - when you tell hadoop where to store files (and temp space), can you
also limit how much is used?  That is, we can create one or more
directories for hadoop to use, but sometimes those directories will be
on filesystems that are shared with other uses, and we'll need to limit
how much hadoop can consume.

  - for machines with multiple cpu cores, can you control how many jobs
are run simultaneously?  For example, on desktops it would be nice to
make sure that only one hadoop job is running at a time so that the
machine stays responsive to the user.

  - similar question regarding RAM.  Some machines have more or less
RAm than others.  Can we restrict how much RAM is available to the hadoop
jobs?


Thanks a lot for your help

Reply via email to