Browse Prior Art Database

Dynamic Load Sharing for Distributed Computing Environment

IP.com Disclosure Number: IPCOM000116042D
Original Publication Date: 1995-Jul-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 6 page(s) / 280K

Publishing Venue

IBM

Related People

Lin, DH: AUTHOR [+2]

Abstract

Disclosed is a dynamic load sharing services based on Open Software Foundation (OSF) Distributed Computing Environment (DCE). Although DCE provides a wide spectrum of services for developing distributed application, it does not provide load sharing services. This disclosure proposes an algorithm to make use of the technique of threshold and a smoothing function to reduce the overhead incurred by workload information update. The algorithm is also designed to take advantage of the services provided by DCE in order to fit in the existing distributed environment.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 22% of the total text.

Dynamic Load Sharing for Distributed Computing Environment

      Disclosed is a dynamic load sharing services based on Open
Software Foundation (OSF) Distributed Computing Environment (DCE).
Although DCE provides a wide spectrum of services for developing
distributed application, it does not provide load sharing services.
This disclosure proposes an algorithm to make use of the technique of
threshold and a smoothing function to reduce the overhead incurred by
workload information update.  The algorithm is also designed to take
advantage of the services provided by DCE in order to fit in the
existing distributed environment.

      An important advantage of distributed system is the potential
of resource sharing to provide the users with a rich collection of
resources that are usually unavailable or highly contended for a
stand alone system.  Among the computing resources, the processing
power is especially interesting.  The computing power of systems is
growing larger and larger.  However, the system users can rarely
utilize the computing power that the system provides.

      To utilize the huge computing resource provided by distributed
system, this disclosure presents a dynamic load sharing algorithm
based on OSF DCE.  The OSF/DCE provides services and tools that
support the creation, use, and maintenance of distributed
applications in a heterogeneous computing environment.  One of the
major goals of DCE is to facilitate resource sharing in the
networking environment.  However, DCE does not provide load sharing
services so that systems can transparently share their processing
power together.  In the current OSF/DCE environment, when a client
requests a service from the name space database (that is, Cell
Directory Service Clearinghouse), in the case of multiple compatible
servers, the RPC runtime randomly returns a compatible server binding
to the client (1).

      In a single-cell environment, if all application server systems
are homogeneous machines with the same configuration (that is, same
CPU power, same amount of RAM, same DASD, and so on.), the current
randomly returned server binding might not cause any inefficient
resources sharing.  However, in the real world, the application
servers are usually on heterogeneous machines and the workload of
each machine varies.  A randomly selected server binding is not
efficient enough to best utilize all the resources in this
environment.

      Workload Characterization - To achieve faster performance for a
job, job is usually assigned to the lowest load server machine.  But
the question is: how is the workload of a certain workstation
determined?  The criteria selected for measuring machine workload in
this disclosure are:

      Processor utilization - Processor usage is usually considered
as the main factor in determining the workload of an individual
system.  Most of the load sharing schemes use queue length of the
scheduler (2,3) to determine the cur...