TY - JOUR
T1 - Hardware-aided update acceleration in a hybrid Semantic Web database system
AU - Heinrich, Dennis
AU - Werner, Stefan
AU - Blochwitz, Christopher
AU - Pionteck, Thilo
AU - Groppe, Sven
N1 - Funding Information:
This work is funded by the German Research Foundation (DFG) project GR 3435\9-1.
Publisher Copyright:
© 2018, Springer Science+Business Media, LLC, part of Springer Nature.
Copyright:
Copyright 2020 Elsevier B.V., All rights reserved.
PY - 2020/10/1
Y1 - 2020/10/1
N2 - In this paper, we focus on update optimizations in a Semantic Web database system aided by a field programmable gate array (FPGA). Many databases utilize B+-tree index structures for querying data. In this scenario, the B+-tree levels are distributed between the host system with the lower inner levels including the leaves and the FPGA with the upper inner levels including the root. In this way we can perform a parallel search inside the nodes of the FPGA by exploiting its parallel nature. Since update operations presuppose a search for the correct position inside a B+-tree leaf, these operations can benefit from these parallel searches. We present our scheduler ideas to estimate the expected benefit against the setup of the system and further adjustments made necessary by performed updates. In a best, average and worst-case scenario, we show how our scheduler would calculate the possible acceleration of such a system.
AB - In this paper, we focus on update optimizations in a Semantic Web database system aided by a field programmable gate array (FPGA). Many databases utilize B+-tree index structures for querying data. In this scenario, the B+-tree levels are distributed between the host system with the lower inner levels including the leaves and the FPGA with the upper inner levels including the root. In this way we can perform a parallel search inside the nodes of the FPGA by exploiting its parallel nature. Since update operations presuppose a search for the correct position inside a B+-tree leaf, these operations can benefit from these parallel searches. We present our scheduler ideas to estimate the expected benefit against the setup of the system and further adjustments made necessary by performed updates. In a best, average and worst-case scenario, we show how our scheduler would calculate the possible acceleration of such a system.
UR - http://www.scopus.com/inward/record.url?scp=85048748900&partnerID=8YFLogxK
U2 - 10.1007/s11227-018-2462-y
DO - 10.1007/s11227-018-2462-y
M3 - Journal articles
AN - SCOPUS:85048748900
VL - 76
SP - 7961
EP - 7984
JO - Journal of Supercomputing
JF - Journal of Supercomputing
SN - 0920-8542
IS - 10
ER -