Friday 11 April 2008
For a client I work for, it is normal procedure to do a security check before servers go into production. This is not obvious for most organisations though.
In the years I spent at various organisations I have never found a situation before where it was policy to do a hardeningscheck and/or a hack test for every single new server that is to be placed in production.
The hardening check consists of checking if various services or daemons are switched off, IPSec is used, no default login accounts are used, a host-based firewall is used, all patches are applied, etc. This is a check on the "inside" of a system.
The hacktest will check the outside. What TCP/UDP ports are open, is the system vulnerable to SQL-injection or cross-site scripting, etc.
These tests should not done by system administrators, but by security professionals (preferably from an external company). The results should be documented in a report with non-compliances and tips. Systems are only allowed in production after these checks are done. This is done for every single system.
If changes are made to the systems, the tests must be repeated.
A risk analysis should be part of the process.
I think more companies should implement this strategy. It takes time, slows down implementations, and costs money. I know. But systems usually will be in production for many years. Making sure the system has a secure start is the least one can do.