Hello! I am currently in the design stages of a project using
ORACLE version 6 RDBMS and SQL*net V6.0.36.1.1 to have DOS 5.0 PC's
do client access (using ORACLE DOS tools) to the database server on a
UNISYS U6000/65 (SVR4) box. The problem is one of security. We'll have
DCA/RLN (Remote Lan Node) access for dial-up customers, who'll be querying
the DB. There will be ORACLE permit security which will prevent users
from doing unauthorized things like deleting from tables they shouldn't.
However, there will also be things which would be easily handled with
stored procedures and triggers (like: update this table with a charge record
whenever that table is selected from) which can only (I believe) be done
in ORACLE 6 via the application logic. What scares me is a user who fires
up SQL*Plus as a DOS tool and then connects to the DB; since they don't
really log on to UNIX there's no UNIX security, and the database making
a SQL*net connection can't tell if it's an application or an interactive
interface connected to it. Thus, they might do things which the application
must be able to (like insert rows in a table) without the application-based
integrity constraints. The best thing I can think of so far is to have them
enter an application password, which is decrypted by a C program; it would
then connect to a the database with a password which never appears as a file
on the PC. However, a good hacker could dissassemble the encryption
algorithm, leaving me exposed. Anybody have an experience with this?
Thanks in advance!
Andrew Jones,
New Brunswick Geographic Information Corporation,
Fredericton, New Brunswick, Canada
E-mail: LRPR _at_ UNB.CA