Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Distributed Hierarchical Naming of Tasks

IP.com Disclosure Number: IPCOM000100436D
Original Publication Date: 1990-Apr-01
Included in the Prior Art Database: 2005-Mar-15
Document File: 2 page(s) / 66K

Publishing Venue

IBM

Related People

Cohn, DL: AUTHOR [+2]

Abstract

Distributed processes consist of autonomous, cooperating tasks which can be identified by task name. This article addresses the distributed assignment of unique, location-independent names to tasks in a distributed computing environment. The naming mechanism is fully distributed; thus, it is not subject the single point of failure problem. Furthermore, the names assigned are location independent. This feature isolates task names from network configuration complexities and makes task migration possible. Finally, the naming mechanism allows each task to be associated with some accountable entity.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Distributed Hierarchical Naming of Tasks

       Distributed processes consist of autonomous, cooperating
tasks which can be identified by task name.  This article addresses
the distributed assignment of unique, location-independent names to
tasks in a distributed computing environment.  The naming mechanism
is fully distributed; thus, it is not subject the single point of
failure problem.  Furthermore, the names assigned are location
independent.  This feature isolates task names from network
configuration complexities and makes task migration possible.
Finally, the naming mechanism allows each task to be associated with
some accountable entity.

      The new distributed hierarchical naming method views the
distributed system as a collection or forest, of name trees.  Each
name tree begins at a root task.  By properly restricting the naming
of root and non-root tasks, the decentralized system can support
globally unique names.  In order to accomplish this, each computer in
the interconnection must have a unique machine name.

      Root task names are constructed from two components: a machine
name, indicating the machine on which the root task was created, and
a base name that is unique within the machine.  For example, if a
root task with base name BASE is created on a machine named MACHINE,
its full name would be /MACHINE.BASE.  Before assigning the name, a
name server on MACHINE could ensure that no other local root task was
already named BASE.

      A similar strategy is used to assign globally unique names to
non-root tasks.  Every task is assigne...