Browse Prior Art Database

Automated Accounting Storage Transaction Unit Methodology

IP.com Disclosure Number: IPCOM000110338D
Original Publication Date: 1992-Nov-01
Included in the Prior Art Database: 2005-Mar-25
Document File: 2 page(s) / 58K

Publishing Venue

IBM

Related People

Beardsley, D: AUTHOR [+4]

Abstract

Distributed Processing systems contain disparate operating environments with corresponding disparities in the cost of doing a particular operation. In the distributed environment, automated storage management between many different types of computing technology requires a consistent methodology for the measurement of resource utilization.

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

Automated Accounting Storage Transaction Unit Methodology

       Distributed Processing systems contain disparate
operating environments with corresponding disparities in the cost of
doing a particular operation.  In the distributed environment,
automated storage management between many different types of
computing technology requires a consistent methodology for the
measurement of resource utilization.

      This invention optimally and dynamically defines the cost
metric of a unit of work in an enterprise.  It does this by resolving
the relative cost of the storage media involved, the CPU utilization
required, and the quantity of network resource utilized.  The
algorithm involved is of the form:
       TU = (pQd + pQlp + pQc)k
where:
TU   is the transaction unit,
pQd  is the block cost of a unit of storage (tape, DASD, optical),
pQlp is the aggregate cost of the communications overhead required to
complete the transaction,
pQc  is the relative cost of a unit of work in the given computing
environment (Note: This value heuristically combines the mainframe
cost factor with that of the workstation.), and
k    is a constant, determined automatically based upon environment
for normalization purposes.

      The k constant is derived dynamically based upon a fixed block
(2k) transaction polling sequence which determines the current load
for a given path (e.g., the current utilization of a path to the
archive server consisting of token ring and channel attachments) as a
function of total theoretical bandwidth of a path.  The polling
sequence simply tests the circuit (from to from) time for each path
defined for a client to the archive server.  This value (in
milliseconds) is then divided by the total raw theoretical capability
of the path, with the minimal throughput portion used as the single
metric (e.g. the slowest element is ut...