AnsweredAssumed Answered

why "Run on hadoop" in eclipse runs much faster than export as JAR file to run in terminal for a MapReduce program

Question asked by foolyoghurt on Mar 30, 2013
Latest reply on Mar 30, 2013 by Ted Dunning
When I run a MapReduce program in Eclipse with "Run on Hadoop", it just took **68** seconds.Howerver, as I exported it as JAR file with Eclipse and run it in the linux terminal, it took **199** seconds.And the output were just the same.

I am now so confused about it.Why there is a such differece between them.Does Eclipse do some optimization? And then I did some optimization with changing my hadoop configuration, it turn out to be faster when run in the liunx terminal(<strong>184</strong> seconds).But in Eclipse it remained the same **68** seconds.

So who can tell my why. Any help will be greatly appreciated.