You do not have permission to delete messages in this group
Copy link
Report message
Show original message
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to shiv...@cs.duke.edu, hadoop-...@googlegroups.com
Hello,
I want to know how Starfish computes the shuffle time in the reduce phase.
I my experiment, shuffle time is very high. But in Starfish it is less.
I am computing the shuffle time (from job history file) as
SHUFFLE_FINISHED - REDUCE_START_TIME
However, same value is not there from Starfish. Why this is so?
Thanks,
YoungKun Min
unread,
Jan 8, 2015, 12:18:35 PM1/8/15
Reply to author
Sign in to reply to author
Forward
Sign in to forward
Delete
You do not have permission to delete messages in this group
Copy link
Report message
Show original message
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to hadoop-...@googlegroups.com, shiv...@cs.duke.edu
Hello, Amit
I believe starfish uses performance models in their paper. Did you look at the chapter 6 in this technical paper(http://www.cs.duke.edu/starfish/files/hadoop-models.pdf)? There are too many variables and they are not real implementations, but I believe there are some points of starfish's view.