Thesis OPTIMIZACIÓN DE TRIPLE PATTERN FRAGMENTS UTILIZANDO STATEFUL CONCURRENT CACHE
Loading...
Date
2019-10
Journal Title
Journal ISSN
Volume Title
Program
DEPARTAMENTO DE INFORMÁTICA. INGENIERÍA CIVIL INFORMÁTICA
Campus
Casa Central Valparaíso
Abstract
En esta memoria se realiza una optimización al sistema de bases de datos: Linked Data
Fragments, utilizando una caché que tiene por objetivos principales mantener
consistencia de resultados entre distintas sesiones HTTP con distintos clientes de forma
concurrente, para así disminuir la cantidad de procesamiento que deberá realizar el
servidor en casos donde ya se conoce el resultado en base a la petición del usuario.
El principal resultado de este trabajo es la verificación de la solución propuesta,
obteniendo datos duros sobre su funcionamiento en un conjunto de servidores que se
utilizaron para obtener métricas de los resultados. Además se realizan distintas pruebas
para poder validar correctamente la implementación de caché en distintos escenario:
funcionamiento independiente, impacto en tiempo de CPU e impacto en un sistema
distribuido.
Al lograr validar esta solución se hace un aporte al sistema de bases de datos, ya que logra
disminuir los tiempos de consultas en hasta un 29.6%, lo que se traducirá en menores
costos para clientes y servidores.
In this document we apply an optimization on the database system: Linked Data Fragments, by using a cache with the main objectives of maintaining consistency of results across different concurrent HTTP sessions with different clients, effectively reducing the needed computing on the server on cases when we already know the answer based on an user's request. The main result of this work is the verification of the proposed solution, obtaining data of the process on a cluster of servers, and then analyzed to get final metrics. Also, we perform different benchmarks to correctly validate the cache implementation in different scenarios: independent runs, CPU time impact and distributed system impact. By confirming the solution we made a contribution to the database system, reducing the query time up to a 29.6%, which translates to lower costs for clients and servers.
In this document we apply an optimization on the database system: Linked Data Fragments, by using a cache with the main objectives of maintaining consistency of results across different concurrent HTTP sessions with different clients, effectively reducing the needed computing on the server on cases when we already know the answer based on an user's request. The main result of this work is the verification of the proposed solution, obtaining data of the process on a cluster of servers, and then analyzed to get final metrics. Also, we perform different benchmarks to correctly validate the cache implementation in different scenarios: independent runs, CPU time impact and distributed system impact. By confirming the solution we made a contribution to the database system, reducing the query time up to a 29.6%, which translates to lower costs for clients and servers.
Description
Keywords
LINKED DATA FRAGMENTS, CACHÉ LRU, OPTIMIZACIÓN, BASES DE DATOS