“Quantifying Eventual Consistency with PBS”, a paper written by AMPLab Grad Students Peter Bailis and Shivaram Venkataraman, along with advisors Michael Franklin, Joseph Hellerstein and Ion Stoica, has been published as a “Research Highlight” selection in this month’s Communications of the ACM. Since 2007 the CACM has published selected papers appearing in ACM conferences that represent significant advances in the field. These papers are chosen by an ACM Committee and are typically expanded to appeal to the broader CACM readership. The resulting article is preceded by a “Technical Perspective” written by an expert in the field, in this case, Philip Bernstein, a Distinguished Scientist at Microsoft Research. In his perspective, Dr. Bernstein discusses the traditional tradeoff in data replication between having to read multiple copies and reading potentially stale data. He states, “Until recently, this compromise was done by guesswork […]. The following paper is a breakthrough that removes the guesswork and replaces it by a principled analysis…”
The original PBS paper appeared in the VLDB 2012 conference, where it was chosen for inclusion in the “Best of VLDB 2012” special issue of the VLDB Journal.
The CACM Research Highlight version of the work can be found here.
Phil Bernstein’s perspective on the paper can be found here.