jps@cat.cmu.edu (James Salsman) (07/12/89)
Many universities, businesses, and other institutions have
large numbers of workstation class computers arranged on a
LAN and sometimes with a gateway for a WAN.
These computers can be equipped with a CD-ROM drive for a
suprizingly small amount of money. There is no need to have
the same CD in each computer -- the system can be organized
as a distributed database.
Unfortunatly, distributed databases are slow unless the
index is kept locally, or perhaps on a dedicated "index
server" machine. For a textual database, each entry in the
index would contain pointers to the workstation with the
indexed data, the CD-filename, and the position in the file.
A typical dialog might look like this:
Your Workstation The Index Server Somebody else's Workstation
---------------- ---------------- ---------------------------
---Request for lookup--->
<--list of index enteries---
----------------------------------------request for data--->
<-------actual data--------------------(background process)
Of course, all of this can be scaled up to a global
database, and scaled out to multimedia data (traditional DB,
full, postscript, full-texts, sound, music, clip-art,
anamations, knowledge bases, etc.)
:James Salsman
--
:James P. Salsman (jps@CAT.CMU.EDU)