[HADOOP-6183] improvement in fairscheduler

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Won't Fix
• Affects Version/s: 0.19.1
• Component/s: contrib/hod

Environment

Fedora9

• Assignee: Unassigned
• Reporter: Anjali M
• Created: 10/Aug/09 08:58
• Updated: 18/May/11 22:03
• Resolved: 18/May/11 22:03

Description

I understand the hadoop fair scheduler does the scheduling at a task level. I think in the pool of jobs if we can queue up the jobs based on a backfilling strategy, the delay which long jobs might face in the current strategy, can be reduced fruther. Has anybody tried such a strategy?