Catalog Home Page

Arms: a decentralised naming model for object-based distributed computing systems

Li, Jia (2010) Arms: a decentralised naming model for object-based distributed computing systems. PhD thesis, Murdoch University.

[img]
Preview
PDF - Front Pages
Download (451kB) | Preview
    [img]
    Preview
    PDF - Whole Thesis
    Download (10MB) | Preview

      Abstract

      Entities communicate with one another in distributed computing systems via symbolic names. Implementing such communication requires a naming scheme that dynamically maps these symbolic names to physical nodes and processes. Traditionally, a centralised name server is deployed to perform such translations. However, a collaborative and dynamic environment requires a decentralised naming system due to reasons of efficiency and reliability.

      ARMS (Adaptive, Randomised and Migration-enabled Scheme) is a novel decentralised naming scheme for distributed object-oriented computing systems. A notable feature of ARMS is that it provides direct naming supports for the patterns of object communication and object migration processes to achieve greater performance and scalability in executing object-oriented software within a distributed environment. These supports are driven by three key components: 1) an adaptive locating protocol that exploits the patterns of object communication and explores the best routing path in the face of the changing network conditions, 2) a randomised overlay that is a scalable and flexible substrate for routing name queries, and 3) a hybrid relocation scheme that provides a transparent and efficient means of referencing migrated objects.

      The performance of ARMS has been examined using a number of real world Java-based benchmarking programs. Based on results in this study, ARMS has found to be superior to its structural counterpart – the Chord model because of the adaptive routing protocol and the resilient overlay. Furthermore, ARMS has shown to be superior in a number of other performance metrics.

      Publication Type: Thesis (PhD)
      Murdoch Affiliation: School of Information Technology
      Supervisor: Fung, Lance, Myers, D. and Wong, Kevin
      URI: http://researchrepository.murdoch.edu.au/id/eprint/5122
      Item Control Page

      Downloads

      Downloads per month over past year