Load Unbalancing to Improve Performance Under Autocorrelated Traffic
Size-based policies have been known to successfully balance load and improve performance in homogeneous cluster environments where a dispatcher assigns a job to a server strictly based on the job size. The authors first examine how size-based policies can provide service differentiation and complement admission control and/or priority scheduling policies. They find that under auto-correlated arrivals the effectiveness of size-based policies quickly deteriorates. They propose a two-step resource allocation policy that makes resource assignment decisions based on the following principles.