• DocumentCode
    2855974
  • Title

    A Hybrid Distributed Garbage Collection of Active Objects

  • Author

    Ning, Zhang ; Zhang, Chunyu ; Xia, Yang ; Xiong, Guangze

  • Author_Institution
    Sch. of Comput. Sci. & Eng., Univ. of Electron. Sci. & Technol. of China, Chengdu
  • fYear
    2008
  • fDate
    29-31 July 2008
  • Firstpage
    13
  • Lastpage
    17
  • Abstract
    Different from traditional garbage collection (GC), in actor system GC identifying whether the object is garbage or not depends on not only its references but also its processing state. It makes garbage collection more complicate. In previous work tracing algorithms such as mark-sweep are utilized which lead to GC too complicated to be suitable for large-scale distributed system. We present in this paper a hybrid distributed garbage collection which is comprised of local GC and global GC. A reverse reference is added for each reference which is from an actor that can be activated (include remotely referenced actors). Then we can utilize tracing algorithm to collect local garbage. We use timestamp algorithm and backward inquiry to identify and collect global garbage. This GC scheme can collect all garbage correctly. Furthermore, it is fault tolerant and suitable for large-scale distributed system.
  • Keywords
    distributed processing; fault tolerant computing; storage management; active object; backward inquiry; fault tolerance; garbage collection; large-scale distributed system; timestamp algorithm; work tracing algorithm; Computer science; Conferences; Embedded software; Fault tolerance; Fault tolerant systems; Large-scale systems; Memory management; Object detection; Runtime environment; Yarn; Distributed Garbage Collection; Hybrid; Timestamp; Tracing algorithm;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Embedded Software and Systems Symposia, 2008. ICESS Symposia '08. International Conference on
  • Conference_Location
    Sichuan
  • Print_ISBN
    978-0-7695-3288-2
  • Type

    conf

  • DOI
    10.1109/ICESS.Symposia.2008.63
  • Filename
    4627123