RE: [SAGE] INCOSE

["Stephen P. Schaefer" <sschaefer@acm.org>]

So of course my curiosity was piqued, and I visited their web site.
Most of their description of systems engineering seemed applicable to
what computer system administrators do.  It appears that we would
constitute a particular specialty.  However, on their page

http://www.incose.org/practice/fellowsconsensus.aspx

was this:

Commentary by Brian Mar
[***Stephen's comments interspersed like so****]
        
        Most systems engineers accept the following basic core
        concepts: 
        
             1. Understand the whole problem before you try to solve it
                [****UH OH.  Computer system administrators rarely have
                the resources to do that.  Indeed, understanding the
                problem is often the majority of the cost of solving
                it.****]
             2. Translate the problem into measurable requirements 
             3. Examine all feasible alternatives before selecting a
                solution [****There are usually so many "feasible"
                solutions that such an exhaustive examination is not
                cost effective***]
             4. Make sure you consider the total system life cycle. The
                birth to death concept extends to maintenance,
                replacement and decommission. If these are not
                considered in the other tasks, major life cycle costs
                can be ignored. 
             5. Make sure to test the total system before delivering
                it. 
             6. Document everything. 

[***back to Stephen***]
That looks representative of their applicability: much of it applies
(e.g., "Document everything."), but significant pieces don't.

Also, part of the systems engineering approach is to model solutions
before building them, but in the case of computer system administration,
it is often more expensive to model a system (representation +
instrumentation + interpretation + measurement) than to build it.  I do
hope that virtualization techniques (e.g., Solaris domains, VMware, Xen)
will begin to give us the tools to do cost-effective modeling.

    - Stephen