Users of big data are often not computer scientists. On the other hand, it is nontrivial for even experts to optimize performance of big data applications because there are so many decisions to make. For example, users have to first choose from many different big data systems such as those dealing with structured data (e.g., Apache Hbase, Mongo DB, Apache Hive, Apache Accumulo, Presto, Spark SQL), graph data (e.g., Pregel, Giraph, GraphX, GraphLab), and streaming data (e.g., Apache Storm, Apache Heron, Apache Flink, Samza). In addition, there are numerous parameters to tune to optimize performance of a specific system. To make things more complex, users may worry about not only response time or throughput, but also quality of results, monetary cost, security and privacy, and energy efficiency.

In more traditional relational databases these complexities are handled by query optimizer and other automatic tuning tools (e.g., index selection tools) and there are benchmarks to compare performance of different products. Such tools are not available for big data environment and the problem is probably more complicated than the problem for traditional relational databases. The aim of this workshop is to bring researchers and practitioners together to better understand the problems of optimization and performance tuning in a big data environment, to propose new approaches to address such problems, and to develop related benchmarks, tools and best practices.