Even a limited theory of system administration should cover some
key aspects of the problem:
- Policy determination,
- Strategic decisions about resource usage,
- Productivity considerations (the economics of the system),
- Empirical verification of strategies and policies,
- Efficiency of policy and of policy implementation,
- Efficiency of the system in doing its job.
More pragmatic details such as the need for software installation and
upgrade have to be tackled at an abstract level, in terms of productivity,
probability of failure, resource usage and so on. Software bugs can be
addressed in terms of productivity or security. Security, in turn can be
viewed as a contest for resources at the level of the system.
The benefits of automation versus human incursion are often discussed
in system administration, sometimes as a bone of contention. This is
one area that a theory of system administration can address objectively
and have a real prospect of answering once and for all. An aspect of this
will be discussed later as an example.
Mark Burgess
2000-03-24