Multiple Resource Scheduling

Multiple Resource Scheduling:

Dominant Resource Fairness: Fair Allocation of Multiple Resource Types
Ali Ghodsi, Matei Zaharia, Benjamin Hindman, Andy Konwinski, Scott Shenker, Ion Stoica
University of California, Berkeley
NSDI, 2011

Max-min fairness on dominate resource share.
Motivated by data center Hadoop/Dyrad execution.
Max-min fairness satisfies four important properties:
Share-incentive, strategy-proofness, envy-freeness, pareto efficiency

My comment:
sharing between non-collaborating, competitive users.

Multi-resource fair queueing for packet processing

UC-Berkeley

SIGCOMM, 2012

They implemented DRF in a time-sharing manner (instead of space sharing), mostly motivated by mulitple resource sheduling in Middleboxes. 

Performance Isolation and Fairness for multi-tenant cloud storage
David Shue, Michael J. Freedman, and Anees Shaikh.

OSDI, 2012

   
System level max-min fairness at the key-value store level.
Assume well-provisioned network, do not deal with in-network resource sharing.  
They mainly deal with: where to place datasets, where to replicate datasets, how to allocate weights to each node, and how to do faire queuing locally to ensure DRF even though some datasets are more (dynamically) popular than others. (They use DRF so they do deal with multiple resources, namely bytes received, bytes sent, and requests. bytes bound by network/storage, requests bound by interrupts???)