MASCARA-FPGA cooperation model: Query Trimming through accelerators

Van Long Nguyen Huu, Laurent d'Orazio, Emmanuel Casseau, Julien Lallet
2021 33rd International Conference on Scientific and Statistical Database Management  
Processing sequences of online queries with an efficient response time is crucial for the new data management systems. To address this challenge, we proposed the ModulAr Semantic CAching fRAmework (MASCARA), in which Semantic Caching (SC) is used for a fast and accurate logs processing tool in a security monitoring system. Although MASCARA pointed out the expensive modules, we have not yet discussed an architecture in which they leverage the accelerators of Field Programmable Gate Array (FPGA).
more » ... Therefore, in this paper, we propose a MASCARA-FPGA cooperation model and related major accelerators. We also explain the pipeline model in which multiple accelerators can run in parallel. Because Query Trimming can reduce the response time by factors of up to 2.8x for a single kernel, we confirm the potential of building the full MASCARA-FPGA model.
doi:10.1145/3468791.3468795 fatcat:vl5coxzkava7fi6lj7n7cnnpla