On Wed Jan 14, 2015 at 12:06:56 +0530, Gayathri Nagarajan wrote:
How to show that an application running on l4re with fiasco.oc is more secure than running the same application on normal operating system?
There is no easy way to do that. One common approach is to look at the amount of source code required for an application, and all the code this application requires to run. Given that most software written by humans contains bugs one can assume that the amount of bugs is bigger with a larger software base than with a smaller one. So having a small software base is generally considered to be a good thing security wise (small trusted computing base (TCB)).
Which application can be chosen for proving the security aspects of l4re? Some of which that I know are
- DMS - Denial of mitigation system
- IDS - Intrusion Detection system
- Firewall
etc...
But I want a very basic and simple application to demonstrate the security when compared with running the same on normal os.
One of the key aspects here is that applications can be split into security-critical parts and non-critical parts. Typically the first part is much smaller than the other. This smaller part can then be put into a microkernel application and separated from the big application or even from the whole guest OS. This tremendously reduces the TCB for the security-critical part. Please look at http://os.inf.tu-dresden.de/papers_ps/eurosys2006.pdf for a far more elaborate explanation.
Adam