Hi Harsh,
i thank you for your fast reply. Yes i want to implement something similiar to the paper you sent me. But i have still some problems with the structure of the projects in svn. You wrote that i should work on MR2. That means YARN doesn't it? After going to http://svn.apache.org/repos/asf/hadoop/common/trunk/ i see there a lot of projects. According to what you wrote me i think i should work on http://svn.apache.org/repos/asf/hadoop/common/trunk/hadoop-yarn-project/? Is that true? But even there (http://svn.apache.org/repos/asf/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/) exists many sub-projects. Could you please explain me which project is the right one to work on and how can i import the projects in eclipse as normal eclipse projects? Thank you > From: [email protected] > Date: Wed, 15 May 2013 16:24:12 +0530 > Subject: Re: which mapreduce project to work on? > To: [email protected] > > Are you looking to do something similar to > http://www.neilconway.org/docs/nsdi2010_hop.pdf? > > Ideally you should be working on MR2 which replaces MR1, and MR2 is > available in the current 2.0.x releases as well as in trunk. If not a > specific revision of trunk, you can pick the 2.x release (or branch-2) > if you seek some stability while developing your work - merging to > trunk is easier from there. > > On Wed, May 15, 2013 at 3:13 PM, Elma Barjami <[email protected]> wrote: > > Hi together, > > > > I'm new in Hadoop and I need some help. > > What I'm trying to do is to modify the MapReduce framework , that some > > reduce work begin before all maps ends their job. > > > > The very first question is which project from svn is more adequate to work > > on/develop? > > > > Second, is there a possibility to implement such a thing and if yes which > > classes/packages should I look in? > > > > > > thanks, > > Elma > > > > > > > > -- > Harsh J
