Yuehai Xu , Wayne State University Facebook, Detroit Menlo Park
Eitan Frachtenberg , Facebook Eitan Frachtenberg, Menlo Park Palo Alto
Song Jiang , Wayne State University, Detroit
Mike Palecezny , Facebook, Menlo Park
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/MIC.2013.80
Key-value caches play an important role in the infrastructure of many Web sites. Analyzing their workloads can lead not only to a better understanding of how these unique tools are used in production, but can also lead to better design decisions in future implementations. Abstract This paper analyzes the workload of Memcached at Facebook, one of the world's largest key-value deployments. We look at server-side performance, request composition, caching efficacy, and key locality. Our observations lead to several design insights and new research directions for key value caches, such as the relative inadequacy of the least-recently-used replacement policy.
Internet, Reactive power, Sun
Yuehai Xu, Eitan Frachtenberg, Song Jiang, Mike Palecezny, "Characterizing Facebook's Memcached Workload", IEEE Internet Computing, , no. 1, pp. 1, PrePrints PrePrints, doi:10.1109/MIC.2013.80