Mailing-List: contact bigtop-dev-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: bigtop-dev@incubator.apache.org
Received-SPF: pass (athena.apache.org: domain of shaposhnik@gmail.com
 designates 209.85.214.47 as permitted sender)
MIME-Version: 1.0
Sender: shaposhnik@gmail.com
Date: Fri, 4 May 2012 10:50:37 -0700
Message-ID: 
 <CA+ULb+v+PhDwAza6m82KFTSP45QTfXYcjA7Qy8Jqs8Tc8+mBSg@mail.gmail.com>
Subject: Hama in Bigtop
From: Roman Shaposhnik <rvs@apache.org>
To: bigtop-dev@incubator.apache.org
Cc: edwardyoon@apache.org
Content-Type: text/plain; charset=ISO-8859-1

I'm really happy to see that Haman integration into Bigtop is making
very steady progress. In fact today, for the first time, I was able
to build packages for Hama and play with it (thanks to Victor!).

That, of course, made me curious about a couple of things wrt.
to Hama deployment/usage. I'm CCign Edward for his advice
(Edward, feel free to CC hama mailing list if you think its appropriate).

So, here's what I'm curious about:
   * What does Hama *really* depend on in Hadoop? It seems that
      given its architecture the only dependency should be on HDFS
      (since it essentially replaces Mapreduce layer) and common.
   * I see that there's a way of running hama's own Zookeeper service.
      Is there a way to utilize an existing instance of Zookeeper?
   * Hama on YARN -- does it still need its own services running (BSPmater,
     Groom) or does it become a 100% the same type of application that
     mapreduce is on top of YARN ?
   * What's the best way to run as many 'example' BSP jobs as possible?
      Is there an equivalent of hadoop-examples.jar ?

Thanks,
Roman.