Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 16 views

    Shared Memory

    Uploaded by

    Harsh Nebhwani
    The document discusses distributed shared memory which provides a virtual address space shared among computers in a distributed system without physically shared memory. It provides advantages like simplified programming through read/write abstraction and avoiding data movement across address spaces. Challenges include ensuring replica consistency and overhead from asynchronous synchronization when implementing as message passing.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Shared Memory

    Uploaded by

    Harsh Nebhwani
    16 views3 pages
    The document discusses distributed shared memory which provides a virtual address space shared among computers in a distributed system without physically shared memory. It provides advantages like simplified programming through read/write abstraction and avoiding data movement across address spaces. Challenges include ensuring replica consistency and overhead from asynchronous synchronization when implementing as message passing.

    Original Title

    shared memory

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document discusses distributed shared memory which provides a virtual address space shared among computers in a distributed system without physically shared memory. It provides advantages like simplified programming through read/write abstraction and avoiding data movement across address spaces. Challenges include ensuring replica consistency and overhead from asynchronous synchronization when implementing as message passing.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    16 views3 pages

    Shared Memory

    Uploaded by

    Harsh Nebhwani
    The document discusses distributed shared memory which provides a virtual address space shared among computers in a distributed system without physically shared memory. It provides advantages like simplified programming through read/write abstraction and avoiding data movement across address spaces. Challenges include ensuring replica consistency and overhead from asynchronous synchronization when implementing as message passing.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 3

    ROHINI COLLEGE OF ENGINEERING & TECHNOLOGY

    DISTRIBUTED SHARED MEMORY

    Abstraction and its advantages

    Distributed Shared Memory is a resource management component of a distributed


    operating system that implements the shared memory model in distributed systems, which
    have no physically shared memory. The shared memory model provides a virtual address
    space that is shared among all computers in a distributed system.

     It is an abstraction provided to the programmer of a distributed system.

     It gives the impression of a single memory. Programmers access the data across the
    network using only read and write primitives.

     Programmers do not have to deal with send and receive communication primitives
    and the ensuing complexity of dealing explicitly with synchronization and
    consistency in the message passing model.
     A part of each computer’s memory is earmarked for shared space, and the
    remainder is private memory.

     To provide programmers with the illusion of a single shared address space, a


    memory mapping management layer is required to manage the shared virtual
    memory space.

    Fig : Abstract view of Distributed Shared Memory Advantages of DSM

     Communication across the network is achieved by the read/write abstraction that


    simplifies the task of programmers.

    CS8603 DISTRIBUTED SYSTEMS


    ROHINI COLLEGE OF ENGINEERING & TECHNOLOGY

     A single address space is provided, thereby providing the possibility of avoiding


    data movement across multiple address spaces, and simplifying passing-by- reference
    and passing complex data structures containing pointers.

     If a block of data needs to be moved, the system can exploit locality of reference to
    reduce the communication overhead.

     DSM is economical than using dedicated multiprocessor systems, because it uses


    simpler software interfaces and off-the-shelf hardware.

     There is no bottleneck presented by a single memory access bus.

     DSM effectively provides a large (virtual) main memory.

     DSM provides portability of programs written using DSM. This portability arises
    due to a common DSM programming interface, which is independent of the
    operating system and other low-level system characteristics

     When multiple processors wish to access the same data object, a decision about how
    to handle concurrent accesses needs to be made. If concurrent access is permitted by
    different processors to different replicas, the problem of replica consistency needs to
    be addressed.

    Challenges in implementing replica coherency in DSM systems

    1. Programmers are aware of the availability of replica consistency models and from
    coding their distributed applications according to the semantics of these models.

    2. As DSM is implemented as asynchronous message passing, it faces the overhead of


    asynchronous synchronization.

    3. Since the control is given to memory management, the programmers lose the ability
    to use their own message-passing solutions for accessing shared objects.

    Issues in designing a DSM system:


     Determining the semantics to allow for concurrent access to shared objects.

    CS8603 DISTRIBUTED SYSTEMS


    ROHINI COLLEGE OF ENGINEERING & TECHNOLOGY

     Determining the best way to implement the semantics of concurrent access to


    shared data either to use read or write replication.

     Selecting the locations for replication to optimize efficiency from the system’s
    viewpoint.

     Determining the location of remote data that the application needs to access, if full
    replication is not used.

     Reducing communication delays and the number of messages that are involved under the
    covers while implementing the semantics of concurrent access to shareddata.

    CS8603 DISTRIBUTED SYSTEMS

    You might also like