Accelerating Join Operation for Relational Databases with FPGAs

Abstract

In this paper, we investigate the use of field programmable gate arrays (FPGAs) to accelerate relational joins. Relational join is one of the most CPU-intensive, yet commonly used, database operations. Hashing can be used to reduce the time complexity from quadratic (naïve) to linear time. However, doing so can introduce false positives to the results which must be resolved. We present a hash-join engine on FPGA that performs hashing, conflict resolution, and joining on a PCIe-attached system, achieving greater than 11x speedup over software.

GPU Acceleration for SQL Queries on Large-Scale Distributed Systems

“…Regardless of the specifics, SQL remains highly applicable in the Big Data era thanks to the power of its declarative syntax. As such, significant efforts have been made in improving the performance of SQL, targeting parallel heterogeneous hardware architectures, which incorporates many processor types in a single machine [5,6,7,8,11,12,13,14,16].…”

An empirical evaluation of High-Level Synthesis languages and tools for database acceleration

“…Obtaining an optimal scheduling in BSV can sometimes require some extra effort from the designer, due to the strict sequentially consistent paradigm of the language, which might not be obvious to designers with module mkSorter (SortBox#(n,t) [2], &temp [3]); compAndSwap(&temp [4], &temp [5]); compAndSwap(&temp [6], &temp [7]); ... an RTL background. Although Bluespec adopts many syntax expressions from SystemVerilog, the learning curve is steep even for developers with background in hardware design.…”

“…Mueller et al [3] presented sorting networks and the same median operator with sliding window that we use in this work. Halstead et al [4] accelerate join operations using an FPGA, where columns with identical values are chained together in an address table and the hash probe is done in parallel. Istvan et al present an FPGAbased hash table design with chaining and parallel lookups for a memcached server [33].…”

“…For aggregation, we implemented the median operator with a sliding-window, also used in [3]. Finally, for table joins, we included a hash probe algorithm to accelerate hash join operations [4].…”

“…Processing large volumes of data at high speed is of great interest to companies and organisations. Database queries often make exhaustive use of a significant set of algorithms, some of which are good candidates for FPGA acceleration, in order to substantially improve overall query processing performance [1,2,3,4].…”