From my experience...the systems science!

From all the experiences, I would say many times system administration is a combination of -- scientific processes, philosophy and engineering. And more of scientific process in many aspects, a computer science by no means mutually exclusive.

First, it does have a series of methodologies and best practices that are often outlined in the few good books on system administration out there. Not that they are always followed by any means or that they fit every situation, but a large amount of trial-and-error in doing things like setting up machines and preparing for outages went into setting up those practices.

Second, even when handling issues when they arise, system administrators often have to use "scientific" principles and methodologies (such as logic, deduction, inference, etc) in order to deal with them. Research online and off on the subject matter, attempts to solve problems through trial-and-error, gathering empirical evidence from such attempts to solve a problem or improve a process with different methods--those are all fairly scientific if you ask me.

Yes, at times, there does seem to be an almost "black art" feel to system administration. There is rarely ever a single solution to any problem nor a single methodology that works every time. Moreover, at times it is possible to recognize the work/thought processes of some system administrators just as you would the creation and end result of a painting by Picasso or of a musical piece by Chopin.

For me, when setting up new systems and architectures as well as when troubleshooting problems as they arise both generally have proven processes that can work to get the job done. At the very least, those processes and mental or physical checklists often help immensely by guiding me down a path that might work, or at least from which I can learn where it needs to be changed/improved in order to make it work. I may need to take some creative leaps in order to get to there from here, but underlying it all are the "scientific" principles such as logic and deduction mentioned above.

At its very core, however, isn't one large facet of "scientific endeavor" really the solving of problems along with developing the methodologies to do so? And setting up systems to perform certain tasks as well as troubleshooting or planning for issues that arise or that might arise are also problems that need to be solved. In some generic cases, solutions already exist and their processes can be easily followed. In other cases, a little more creativity is involved but it is still, at the core, problem solving using "scientific" methods.

Finally Philosophy comes into play when a systems organization employs different components requiring understanding the infrastructure landscape.


Cheers
Manju