FPGA-Based Dynamically Reconfigurable SQL Query Processing

Abstract

In this article, we propose an FPGA-based SQL query processing approach exploiting the capabilities of partial dynamic reconfiguration of modern FPGAs. After the analysis of an incoming query, a query-specific hardware processing unit is generated on the fly and loaded on the FPGA for immediate query execution. For each query, a specialized hardware accelerator pipeline is composed and configured on the FPGA from a set of presynthesized hardware modules. These partially reconfigurable hardware modules are gathered in a library covering all major SQL operations like restrictions and aggregations, as well as more complex operations such as joins and sorts. Moreover, this holistic query processing approach in hardware supports different data processing strategies including row- as column-wise data processing in order to optimize data communication and processing. This article gives an overview of the proposed query processing methodology and the corresponding library of modules. Additionally, a performance analysis is introduced that is able to estimate the processing time of a query for different processing strategies and different communication and processing architecture configurations. With the help of this performance analysis, architectural bottlenecks may be exposed and future optimized architectures, besides the two prototypes presented here, may be determined.

Toward FPGA-Based Semantic Caching for Accelerating Data Analysis with Spark and HDFS

“…While the software section is composed of a Query analyzer, Spark SQL and the Hadoop Distributed File System (HDFS). This architecture could be later improved by introducing partial reconfiguration to the FPGA architecture to have the ability to do as much queries as possible by the query engine like what was done in [8], [18]. This architecture could also be improved by adding more than one FPGA in the architecture to have different FPGAs executing different queries at the same time.…”

“…Static approaches which use FPGAs for query processing [4], [8], [18] succeeded in implementing query engines capable of doing SQL based processing (select, project join, etc.) with a very high throughput.…”

“…Although some of these existing solutions handled queries for large volumes of data (100 TB in [18]), to the best of our knowledge, the proposed FPGA architectures cannot process huge data stored in multiple servers by the use of a distributed file system. The architecture that we propose, using the Distributed File Systems HDFS, massively parallel processing on Spark SQL, hardware acceleration and semantic caching in FPGA could be the solution to the problem of executing queries in Big Data in a relatively short time.…”

“…HDFS and Spark rely on GPPs and consequently inherit from GPP limitations mentioned previously. Regarding hardware acceleration, [4], [8], [14], [15], [18] have shown that Field Programmable Gate Arrays (FPGAs) could be a better alternative to GPPs in the field of database applications in general and for query-execution purposes with an increasing data throughput with lower energy effort. But not all queries can be easily implemented and processed in a FPGA.…”

AxleDB: A novel programmable query processing platform on FPGA

“…Previous studies have looked into designing efficient query processing engines, employing vector architectures [38], ASICs [4], GPUs [39] or hybrid [40], [41]. On the other hand, other approaches either used FPGAs statically [2], [24], or they leveraged dynamic reconfiguration to better fit the requirements of each query [42], [43], [44]. We differ from these works by following a static but programmable approach in query processing and data management, as we can support as many operations as we need, without requiring runtime reconfiguration.…”