On Thu, Mar 14, 2013 at 11:50 AM, Konstantin Boudnik <[email protected]> wrote: > Hadoop 1.x based stacks are destined to be supported in 0.3.x line of BigTop > releases. 0.3.1 is currently in - slow - progress. There are few outstanding > JIRAs on that and you're very welcome to help with it!
To add to what Cos said: Bigtop is totally community and consensus driven. If there's a bunch of folks who'd be interested in a Bigdata management distribution based on Hadoop 1.X -- they are more than welcome to congregate around branch-0.3 and do periodic releases from there. Personally, I believe that Hadoop's 2.X HDFS is a better choice at this point then Hadoop 1.X's. MapReduce is a different matter -- arguably it is still in alpha stage but there seems to be a lot of interest in investing into stabilization of it. Hence most of the folks involved in Bigtop so far feel like overall Hadoop 2.X may be a more interesting platform. That's why you see Bigtop releases based on it. Hope this helps. Thanks, Roman. P.S. Of course, in the ideal world, HDFS and YARN and frameworks like Mapreduce would be separate projects and you'd be able to mix and match them freely. Alas -- we're not quite there yet :-(
