hadoop-mapreduce-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ravi Prakash <ravi...@ymail.com>
Subject Re: Doubt regarding hadoop simulator
Date Tue, 17 Sep 2013 15:09:30 GMT
Suresh!

Rumen is used to generate a trace file from the job history files on a pre-existing cluster.
This trace file can then be fed into gridmix (for example) to "simulate" the same workload
on that cluster again (or another cluster for that matter). https://hadoop.apache.org/docs/stable/rumen.html.
Rumen also allows you to specify a scaling factor, so if you were able to get a trace for
a much bigger cluster, you could scale it down to run on a much smaller cluster.

I'm afraid I'm not familiar with Mumak. 

You might also be interested in this JIRA: https://issues.apache.org/jira/browse/YARN-1021
. Unfortunately, it hasn't been checked into the repository yet, so you will have to apply
the patch yourself.

HTH
Ravi





________________________________
 From: Suresh S <sureshhot@gmail.com>
To: mapreduce-dev@hadoop.apache.org 
Sent: Tuesday, September 17, 2013 12:48 AM
Subject: Doubt regarding hadoop simulator
 

Hello,

     I am searching for MapReduce simulator in online.
I heared some names like *Rumen and Mumak.*
**
But unable to understand thease simulator.

I have modified some changes on Fair scheduling.
i want to run the simulation for same workload for original Fair scheduler
and modified fair scheduler. And see the difference in the response time,
fairness, locality and network traffic.

I dont have real enviornment to run my experiments.

Please help me in this regard. Is is possible with rumen and mumak?
Is there any other simulater available?

Thanks in Advance...

*Regards*
*S.Suresh,*
*Research Scholar,*
*Department of Computer Applications,*
*National Institute of Technology,*
*Tiruchirappalli - 620015.*
*+91-9941506562*
Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message