Having seen both these paradigms "from the inside" I can make two vague general comments. There is not going to be one best paradigm for distributed computing. Some paradigms are better for some problems.
That said, MapReduce has always been really for one type of problem: analytics / log processing. It's great for anything shaped like that. It's amazing how it's been abused to do more. But it's weird for a general paradigm. Hence I generally find these newer paradigms that are emerging better, since at least they're more purpose-built for general purposes. Hadoop's what we've got easy access to in 2012. I think that will start to change in 2013. Sean On Fri, Apr 20, 2012 at 6:37 PM, Mike Spreitzer <[email protected]> wrote: > I am trying to understand the comparative virtues of map-reduce vs. the > programming model found in Pregel. Both cite BSP as inspiration, but > Pregel's model includes iteration, per-component state, and selective > activation --- all of which are absent in map-reduce. While one could > implement Pregel (or pretty much anything else) on top of Hadoop with > sufficient additional client code, I am trying to compare the quality of > doing that with instead taking the programming model of Pregel, simplified > to apply to key-value data instead of graphs, as the fundamental > abstraction and building other things (as needed/desired) on top of that. > As noted in other replies, the additional features of the Pregel model > look like they could be very beneficial for solving the problem I posed. I > would like to do an actual comparison. So I want to know the best way to > solve this maintenance problem using map-reduce. > > Thanks, > Mike > > > > From: Saikat Kanjilal <[email protected]> > To: "[email protected]" <[email protected]> > Cc: "[email protected]" <[email protected]> > Date: 04/20/2012 12:21 PM > Subject: Re: shortest-path maintenance > > > > Is it a requirement to use map reduce? Also how does mahout play into > this? Potentially you could build mappers that could reference an in > memory graph and have an API that pre calculates dijkstra or astar up > front. You could then add or remove a node as part of the reduce process > that references this graph and recalculates dijkstra or astar in a closed > feedback loop. However its not obvious to me that mapreduce is the > appropriate tool to do this. > > Some more context into the problem and how mahout fits in would be great. > > Sent from my iPhone > > >
