综合论文：0AComparisonofjoinAlgorithmsforLogProcessinginMapReduce（英文）.pdfVIP

A Comparison of Join Algorithms for Log Processing in MapReduce Spyros Blanas, Jignesh M. Patel Vuk Ercegovac, Jun Rao, Computer Sciences Department Eugene J. Shekita, Yuanyuan Tian University of Wisconsin-Madison IBM Almaden Research Center {sblanas,jignesh}@ {vercego,junrao,shekita,ytian}@ ABSTRACT that MapReduce is a step backwards and ignores many of The MapReduce framework is increasingly being used to the valuable lessons learned in parallel RDBMSs over the analyze large volumes of data. One important type of data last 30 years [18, 25]. Among other things, the critics of MapReduce like to point out its lack of a schema, lack of a analysis done with MapReduce is log processing, in which a click-stream or an event log is filtered, aggregated, or declarative query language, and lack of indexes. mined for patterns. As part of this analysis, the log of- We agree with all of these criticisms. At the same time, we ten needs to be joined with reference data such as informa- believe that MapReduce is not a passing fad, as witnessed tion about users. Although there have been many stud- by the rapidly growing MapReduce community, including ies examining join algorithms in parallel and distributed many of the big Web 2.0 companies like Facebook, Yahoo!, DBMSs, the MapReduce framework is cumbersome for joins. and