On 12/8/2011 3:46 PM, James B. Byrne wrote:
Do many of you implement fine grained security provisions at the DBMS level using GRANT PRIVILEGE ON object TO user/role? My background with DBMS security is mostly in CODASYL shallow network systems where user level access control to datasets and fields was mainly a matter for the application and / or operating system. Therefor the number of user ids known to the DBMS was minimal. The topic of RDBMS security has arisen in a discussion and, lacking evidence of my own, I am curious to discover just how frequently DBMS userids tied to specific individuals are used in production RDBMS based systems. I am also curious to know how often VIEWS are tied to individual user IDs known to the DBMS rather than to shared user IDs known only to an application through a configuration file. This is not really a PG specific question so if anyone wishes to reply privately rather than to the list that is fine with me. Thanks,
This is probably a bad example. Or maybe a good example of how not to do it.
I have a generic user, with create user privileges. When users run my app (its a windows delphi app) they type in their username/password, I try to connect as them, and if it doesnt work I connect as the generic user, create them, and reconnect as them.
I pretty much use the user to see who is connected, and for audit logging. I have the rights and stuff in a user table, and the delphi app does all the rights checking.
On the website side, I always connect as a webuser, who has full read/write. Even though the website is read only. Err.. I do keep stats, so its mostly read only.
-Andy -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
