Most of the current efforts aimed at improving host security rely on the flawed assumption that adequate security can be provided solely in the application layer with existing operating system security features laying below not being altered [23]. A number of well documented examples have shown that support from secure operating systems is paramount to fighting threats posed by modern computing environments [23,30,24]. For example, Linux community has been trying to eliminate the buffer overflow2.1 threat for years by auditing source code of the programs available for the platform. Nonetheless, it was not before the advent of, so called, Openwall [3] Linux kernel patch that the threat was successfully fought. The application layer exclusive security provision approach, as stated in [10], results in a fortress built upon sand.
Most of the security attacks on UNIX systems today rely, irrespective of the actual exploit mechanism employed, on the way in which access control and privilege delegation mechanisms are implemented in the underlying operating system. The model that UNIX systems follow for the purpose is, so called, Discretionary Access Control (DAC) model. Therefore, the DAC model is to blame for large proportion of security breaches in UNIX environments. When first UNIX systems, and the DAC mechanisms, were developed they were perfectly suited for the environment it was envisaged they would serve in. However, the environment most of the existing UNIX systems operate in today is far from anything people could have envisaged more than three decades ago. Research and development in the field of operating system security is constantly failing to meet the pace at which it is being challenged. Unfortunately, many operating system mechanisms, known to be seriously flawed and/or inapt for use in the new environments, such as DAC, are still widely present in the operating system design and implementation.