FlexSplit: A Workload-Aware, Adaptive Load Balancing Strategy for Media Cluster

A number of technology and workload trends motivate us to consider a new request distribution and load balancing strategy for streaming media cluster. First, in emerging media workloads, a significant portion of the content is short and encoded at low bit rates. Additionally, media workloads display a strong temporal and spatial locality. This makes modern servers with gigabytes of main memory well suited to deliver a large fraction of accesses to popular files from memory. Second, a specific characteristic of streaming media workloads is that many clients do not finish playing an entire media file that reflects the browsing nature of a large fraction of client accesses. In this paper, we design and evaluate two novel load-balancing strategies for media server cluster: FlexSplit and FlexSplitLard, that aim to efficiently utilize the combined cluster memory by exploiting specific media workload properties. New strategies “tune” their behavior to reflect media file popularity changes and other dynamics exhibited by media workload over time. Adaptive nature and improved cluster performance make these strategies an attractive choice for handling dynamically changing workloads by media server cluster.