Collaboration with the Linked Data Benchmark Council

Our research group recently collaborated with the Linked Data Benchmark Council (LDBC) organisation. The goal of LDBC, a non-profit organisation, is to define standard benchmarks for graph analytical and graph query systems. LDBC positions itself in a similar role for graph processing systems which the Transaction Processing Performance Council (TPC) has been fulfilling for relational databases since the early 1990s. Standard benchmarks are useful in a number of scenarios: they enable academic researchers to assess the performance of their prototypes and industry practitioners to compare existing database management tools.

The LDBC Social Network Benchmark (SNB) defines a set of experiments, which run various workloads on top of a social network graph. The latest one is the “Business Intelligence”, which defines queries that are challenging from multiple aspects: both evaluating complex graph patterns and evaluating data warehouse-like aggregation-heavy global queries is required.

The research was done in an international collaboration, including UPC Barcelona, TU München, Neo4j, and Oracle Labs. From BME, multiple researchers and students participated: Gábor Szárnyas (research associate at Fault Tolerant Systems Research Group and MTA-BME Lendület Research Group on Cyber-Physical Systems) József Marton (Assistant research fellow at the Department of Telecommunications) and János Benjamin Antal (Master’s student at the Fault-Tolerant System Research Group).

The results of the collaboration will be presented at the GRADES-NDA workshop of the ACM SIGMOD 2018 conference in June 2018. This research was partially supported by the MTA-BME Lendület Research Group on Cyber-Physical Systems. The benchmark set produced in this research can be used to investigate the performance of graph processing tools, starting from server-side graph database systems to streaming engines that process sensor networks of cyber-physical systems.