The article entitled "Similarity aware shuffling for the distributed execution of SQL window functions", authored by Fábio Coelho, José Pereira and Rui Oliveira from INESC TEC, together with Miguel Matos from INESC-ID, was selected for the Best Paper Award of DAIS 2017.

In this work, the researchers developed a mechanism that is able to improve how analytic functions are executed across a distributed database, in other words, in a database where the data is not all stored in the same location. This work focuses on Window Functions, a very flexible class of analytical functions whose usage is currently experience and increased by Big Data analytics.

Distributed databases can scale the processing power by distributing data and load across a group of identical workers. Nevertheless, the data distribution introduces several challenges related with the distribution of data and load in a distributed database.

Therefore, this paper introduces a correlation mechanism that is able to identify and improve how remote database nodes exchange information during analytical execution. The mechanism can achieve bandwidth savings in the order of three times and it is expected to increase as the number of database nodes involved grow.

The DAIS 2017 was held June 19th-21st in Neuchatel, Switzerland, and is one of three conferences that are part of the 12th International Federated Conference on Distributed Computing Techniques (DisCoTec).