[AURORA-1219] Improve preemptor efficiency

• Type: Task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.8.0
• Component/s: Scheduler
• Epic Link: Performance improvements
• Sprint: Twitter Aurora Q2'15 Sprint 1
• Story Points: 5
• Assignee: Maxim Khutornenko
• Reporter: Maxim Khutornenko
• Created: 23/Mar/15 17:46
• Updated: 04/May/15 13:12
• Resolved: 22/Apr/15 00:27

Description

While AURORA-1158 makes the preemptor asynchronous and helps scheduling loop performance, it loses some preemption efficiency due to adapting the original algorithm to work asynchronously:

• Preemptor repeatedly searches for preemption slots with the same ClusterState increasing the probability of victim collisions.
• Slave/offer/resource mappings are redundantly recomputed for every pending task slot search.
• Pending tasks are repeatedly sized up against every slave (as opposed to the other way around) making internal loop optimizations more expensive (i.e. pending task count is expected to be << slave count under normal conditions).