NetNews Usenet Archive 1993 #1/NN_1993_1.iso/spool/comp/object/4741

Internet Message Format | 1993-01-10 | 1.7 KB

Path: sparky!uunet!pipex!warwick!str-ccsun!strath-cs!ikorbu
From: [e-mail] (Inge Korbu CS90)
Newsgroups: comp.object
Subject: Data Redundancy For OO-Systems
Message-ID: <[e-mail]>
Date: 10 Jan 93 13:15:26 GMT
Sender: [e-mail]
Organization: University of Strathclyde
Lines: 25
Nntp-Posting-Host: fleming-07.cs.strath.ac.uk


I would like to have some information, if available at all, on Data
Redundancy For Object-Oriented Systems. By this I mean systems that
use data diversity (redundant objects) to provide a data fault
tolerant system.
Can Fault tolerant techniques can be applied equally to any part of a
system? Traditionally these techniques are applied to both the
hardware and software aspects: modular redundancy, N-version
programming, recovery blocks are well known techniques whose use is in
increase. By contrast data fault tolerant techniques have largely been
ignored. One possible reason for this is that implementations are highly
complex. Fortunately this situation has been greatly improved by
date/object oriented systems which supply many of the required facilities.
For example if an object does not respond correctly to a given
message, ie. an acceptance test fails, another slightly different
class/object could be tried (retry block) to see if that gives an
acceptable answer...

Information about relevant research efforts, literature,
or existing systems or any thoughts at all would be greatly appreciated.

========================================================================
Inge Korbu, Computer Science 4th Year | INTERNET: [e-mail]
18 Cecil Street, Glasgow, G12 8RH     |
========================================================================