I was recently reading an interesting article, “from tasks to assurances: redefining system administration”, by Alva L. Couch, in the April 2010 issue of ;LOGIN:. He makes a lot of good points, mainly that system administration has been defined by tasks, but should really be defined by assurances (a much more abstract concept). This does make a lot more sense, and perhaps will aid me in finally coming up with a succinct (but complete) answer for when people ask me what I do.
It bears mention that my role as an SA probably isn’t the same as many others. Working at a University has its perks - and one is definitely the inclination towards research, tinkering, and trying new things. Another is the extremely tight budget - I’d estimate that only about 25% of what my group supports (software-wise) actually has a vendor behind it. Most of those services have primary admins assigned to them.
The group I work in is composed of 9 people (including myself). We’re responsible for the architecture (specifically authentication, network-side stuff) for residential/dorm networking (ResNet), all of the student computing labs (\~1,000 user stations) as well as printing therein, the University-wide wireless network (from physical installation through support), and a few other services. That doesn’t include all of the usual ancillary stuff - mail, DHCP, web apps, storage, etc. It would probably surprise most corporate IT types that we more or less function as an independent unit - we share certain services, like Nagios monitoring, with other groups, but do most of our work as a single standalone unit.
My own job (since I’m currently the only part-time person in our group) is probably very different from most SAs. I’m the primary admin for only one user-visible service, which is in the process of being phased out (and is very lightly used currently). For the most part, I’d describe myself as a “floating” admin - I’m usually assigned whatever problems come up that are in my knowledge area (or more than the primary admin can cram into a work week), and also do quite a bit of research (mainly suitability analysis of new technologies). I’m also the DR guy, and am in the process of implementing across the board an automated installation, configuration, recovery and backup system for all of our Linux/Unix boxes. As ironic as this may seem, while I’m not the primary admin for any of our major services, if one of the boxes than run them falls over, it’s more or less my responsibility to get it back up and running. Or at least have the plans and procedures for that laid out.
Something that is also quite different from many enterprise shops is our software budget - which is almost non-existent. Except for a few services - some of the stuff used to administer the Windows and Mac lab machines, the printing systems, and the wireless stuff, we’re pretty much exclusively open source. As a result, there’s rarely a vendor to fall back on, and “fixing a problem” can often involve hours of reading source.
As an aside, I once had a phone screen where the interviewer asked me about a difficult problem I recently solved, and what my methodology was. I ran through about 20 steps. When I finished the list, the interviewer asked me, “at what point would you call the vendor for support?” My response: “vendor? what vendor?”
So I’m trying to think up a good answer to what I do, and what the other SAs in my group do. There’s the work in progress so far:
We provide many different services to our users (all University faculty, staff, students, etc.). It’s my job to ensure that those services are as reliable as possible, function correctly, are secure, and work as well as possible. When something does fail, no matter what it is, it’s my job to get it back to normal. And when a new technology emerges that may increase the quality, reliability or security of a service we offer, it’s my job to evaluate it and, if it is found to be worthy, implement it.