Newsgroups: comp.arch
Path: utzoo!utgpu!watserv1!watdragon!watsol.waterloo.edu!tbray
From: tbray@watsol.waterloo.edu (Tim Bray)
Subject: Re: CD-ROM documents (was Paperless Office)
Message-ID: <1990Dec3.220850.18352@watdragon.waterloo.edu>
Sender: daemon@watdragon.waterloo.edu (Owner of Many System Processes)
Organization: University of Waterloo
References: <11191@pt.cs.cmu.edu> <00940487.15804140@KING.ENG.UMD.EDU> <28083@mimsy.umd.edu> <1990Nov29.162726.11411@mozart.amd.com> <11212@charm.UUCP> <2974@crdos1.crd.ge.COM>
Date: Mon, 3 Dec 90 22:08:50 GMT
Lines: 20

There's been discussion of CD-ROM here recently.  Should just point out one
important architectural issue: CDROMs have a seek time of on the order of
500 ms - that's right, half a second.  This means that the selection of
algorithms and data structures available for use on CD-ROM is very highly
constrained.  Also, it means that a CD-ROM is unlikely to be a satisfactory
database access medium for more than one user at a time.

It is unlikely that the access time issue will improve.  The reason CD-ROM's
are used at all is their extreme cheapness, which is achieved by leveraging
off the high-volume manufacturing technology due to the music industry.  Make
any significant changes to the medium or the players to speed it up, and
you lose that leverage, and the cost advantage compared to magnetic or
more conventional optical media.

On the other hand, CD's just CAN'T BE BEAT as an electronic *dissemination*
medium.  I wouldn't be surprised, in a few years, to see all the major
database and software vendors shipping stuff out on CD's; the recipients, of
course, copy the stuff onto a *real* disk to use it...

Cheers, Tim Bray, Open Text Systems
