Towards autonomic consolidation of heterogeneous workloads

Abstract

Multicore architectures have become the standard both for general purpose PCs and for high-end servers. Such platforms provide an environment that invites consolidating several applications within the same processor to fully take advantage of the available parallelism. As multicore processors become the basic building blocks of data centers and cloud computing platforms, the problem of effective consolidation becomes very challenging as interference between collocated workloads may significantly affect their performance. Server virtualization has offered solutions to the problems of software heterogeneity and fault isolation but not to the problem of performance isolation. Consequently, there is a lot of research on resource partitioning that aims at performance isolation on multicores. Resource partitioning has been shown to be costly and may even result in inefficient resource usage, defeating the purpose of consolidation in data center and cloud computing platforms for reducing infrastructure and electricity costs. As such, consolidating applications is often conflicting with performance isolation, especially since performance degradation due to collocation may be difficult to characterize and foresee. © 2011 ACM.