Hybrid tables for speeding-up data accesses in hybrid database management systems

• Joan Guisado-Gámez ^[1] ; Antoni Wolski ^[2] ; Calisto Zuzarte ^[3] ; Josep-Lluís Larriba-Pey ^[1] ; Victor Muntés-Mulero ^[1]
  1. [1] Universitat Politècnica de Catalunya

     Universitat Politècnica de Catalunya

     Barcelona, España

     
  2. [2] IBM Helsinki Laboratory
  3. [3] IBM Toronto Laboratory

  Mostrar afiliaciones +
• Localización: Actas de las Jornadas de Ingeniería del Software y Bases de Datos (JISBD´11), 2011, ISBN 978-84-9749-486-1, págs. 133-134
• Idioma: inglés
• Enlaces
  • Texto completo (pdf)
• Resumen

  • With the introduction of unexpensive memory, new techniques for reducing the access time to data stored in databases have been proposed. On the one hand, buffer pool techniques have been proposed. On the other hand, main memory database management systems (MMDBMS) have been developed. While buffer pool techniques are based on speeding up the performance by copying the most accessed pages in memory, MMDBMSs propose a new paradigm consisting in designing the internal structures to store the whole amount of data into the main memory of the system. Although MMDBMSs are fast on the retrieval of information, they limit the total amount of data to the total amount of memory available. This restriction may be too strict for many databases whose sizes are often hundreds or even thousands of gigabytes. As a reaction, dual database management systems (DDBMS) appeared as a type of database developed with two different engines. One engine is in charge of providing access to those tuples that are stored in disk, while there is another engine in charge of providing access to those tuples that are stored in memory. Note that DDBMSs force tables to be either in-memory or on-disk. Thus, large tables that do not fit in memory must be classified as on-disk preventing them from benefiting from main memory techniques.