Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!rutgers!ames!ucbcad!zen!xanadu!larry
From: larry@xanadu.uucp (Larry Rowe)
Newsgroups: comp.databases
Subject: Re:  Degree 3 consistency
Message-ID: <3208@zen.berkeley.edu>
Date: Thu, 6-Aug-87 12:28:40 EDT
Article-I.D.: zen.3208
Posted: Thu Aug  6 12:28:40 1987
Date-Received: Sat, 8-Aug-87 11:07:37 EDT
References: <1215@smokey.UUCP>
Sender: news@zen.berkeley.edu
Reply-To: larry@xanadu.UUCP (Larry Rowe)
Organization: Postgres Research Group, UC Berkeley
Lines: 21

my understanding of the current state of rti/oracle xact mgt
is the following:

1. oracle offers in-memory buffer versioning of pages. i
think they only offer degree 3 (is this right?).

2. rti offers degree 1, 2, and 3 consistency with degree 3
being the default.

with respect to jim gray's comments about the overhead, i believe
he was refering to anticdotal (spell?) evidence he gathered from
observing the use of IMS by large xact applications.  in essence,
most people used degree 3 because the performance penalty was 
worth the overhead.  one thing to remember is that IMS applications
are programmer developed and highly controlled applications.  in
other words, the application designers severely limit the kinds of 
queries that can be executed.  the availability of SQL and forms-based
end user interefaces makes it possible to ask questions that touch 
data in very different patterns than a typical IMS application would
allow.
	larry