Direct lookup and hash-based metadata placement - impact on architecture, performance and scalability of local and distributed file systems

• Autores: Paul Hermann Lensing
• Directores de la Tesis: Antonio Cortés Roselló (dir. tes.), André Brinkmann (codir. tes.)
• Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2019
• Idioma: español
• Tribunal Calificador de la Tesis: Shadi Ibrahim (presid.), Julita Corbalán González (secret.), Juan Piernas Cánovas (voc.)
• Programa de doctorado: Programa de Doctorado en Arquitectura de Computadores por la Universidad Politécnica de Catalunya
• Materias:
  • Ciencias tecnológicas
    • Tecnología de los ordenadores
      • Arquitectura de ordenadores
      • Diseño de sistemas de calculo
      • Dispositivos de almacenamiento
• Texto completo no disponible (Saber más ...)
• Resumen

  • New challenges to file systems’ metadata performance are imposed by the continuously growing average number of files existing in file systems. The traditional component based lookup approach can become a performance bottleneck for many workloads.

    An alternative to component-based lookup is the direct lookup approach. It uses hash-based metadata placement to enable direct computation of the metadata location and to completely skip component traversal.

    This thesis evaluates the implications of the direct lookup approach on file system architecture, performance and scalability.

    It includes an analysis of file system traces focusing on the frequency and characteristics of hierarchical operations. These properties had not been sufficiently explored in existing literature and are fundamental to the direct lookup design.

    Full direct lookup file systems are implemented and evaluated for both local and distributed scenarios.