here's a scenario that I just spoke to a customer about, and am wondering if i am correct.

The customer is coding a page to allow real time access to an Oracle DB in a secured noc. the data is sensitive, and has to be secure, but must also be accessible thru the web.

I suggested that the Database server be located in an ultra-secure area. Then use a reliable NOC for co-location of the actual web server. (I suggested SecureBSD for the OS)
Then, using PHP create an SSL connection to the DB to get the required data.

The security features to gain access to this particular info may be hard to implement. For instance, the connections would be coming from all over the US over various providers (including Dialup) so IP block is out of the question.

Does this make sense?

Is it viable.. what have I missed??


MIke

What's secureBSD?

Just use what's best, OpenBSD has shown time and time again it's no more secure than any other OS. It's all about the admin.

Just make sure the person writes clean code. It's easy to write crappy and insecure PHP code.

Just use lots of checks, ACL's and whatever else you need to insure your data integrity. If it's read-only access then make sure the database is read-only (or the DB_user).


The security is only as strong as its weakest link and the skills of the programmer.

-davidu

Ok, this makes sense.

ACL is part of the solution.... is an SSL connection Viable?

Originally posted by MikeM
Ok, this makes sense.

ACL is part of the solution.... is an SSL connection Viable?

Viable and recommended.

If the DB is not on the webserver, tunnel your DB connection. Highly advised.

-davidu

Yes, SSL connection is reliable. But do add some more layers of security. e.g. look into PIN code, security key, PGP, encryption, etc. on top of ordinary passwords.

This will help in case one layer is compromised, the other can save.

Try looking at a challenge and response system. Such has SecurIDs, if the data has to be extremely secure and is worth the cost.

If possible, put a frontend in front of the database, maybe SOAP or XMLRPC under HTTP. And keep the database inside private subnet, don't let the database exposed to the Internet. Every clients will communicate through the frontend, not directly to the database.

Thanx all...

All of these solutions seem a viable and advisable solution to the issue. Implementing them may be an issue, but thankfulkly I dont have to worry about it....


Once again thanx.