The 31st International Conference on Distributed Computing Systems (ICDCS 11) presented the Best Paper Award to Rubao Lee (postdoctoral fellow, CSE at Ohio State), Tian Luo (CSE Ph.D. student), Yin Hua (CSE Ph.D. student at Ohio State), Xiaodong Zhang (CSE Chairman at Ohio State), Fusheng Wang (research scientist at Emory University), and Yongqiang He (software engineer at Facebook) to their paper entitled "YSmart: Yet another SQL-to-MapReduce Translator."
MapReduce has become a standard software framework for big data analytics in distributed systems, where system execution of SQL queries is a critical data processing task. An SQL-to-MapReduce translator automatically converts database queries to MapReduce jobs for their execution in distributed systems. Complex SQL query jobs generated by existing translators, such as the one in data warehouse Hive developed in Facebook, and in MapReduce programming environment of Pig developed by Yahoo!, are executed at unacceptable slow speed. The authors of the paper demonstrate that the poor performance of these translators is caused by the framework of one-operation-to-one-job mapping, which does not consider correlations of input data and input keys among the queries. The authors design and implement a correlation-aware SQL-to-MapReduce translator, called YSmart. The translator achieves superior performance compared with the existing translators in Hive and Pig. The translator has been patched in Hive for an adoption, and an independent version of YSmart will be released for public usage.
ICDCS is an annual conference providing a forum for engineers and scientists in academia, industry and government to present their latest research findings in many aspects of distributed and parallel computing. The annual conference this year was held in Minneapolis, Minnesota, June 21-24, 2011.