This recent post talks about a paper from the American Mathematical Society, arguing in favor of using more open source software in mathematics. Traditionally, academia has been open; indeed, the Internet was created (in part) – the network was created – even Usenet was created – to allow researchers to share information back and forth.
If today’s prevailing ideas of patents and making profits off of ideas had been prevalent then, none of these technologies would have been created. It was scientists (including computer scientists, mathematicians, and others) who shared their research, with researchers from one university freely exchanging with researchers at other universities.
In mathematics, we have proprietary software like MATLAB, Mathematica, S-PLUS, and others. However, there are indeed suitable replacements for most everything: there is R (instead of S) and GNU Octave (instead of MATLAB) for example.
The author of the previously mentioned post also mentions SciLab, one I’ve not heard of before.
As a system administrator, statistical data can be gathered from various sources (uptime, disk usage, trends, etc.) and plotted with some of these tools. I’ve seen articles about using R to do such things.
Otherwise, encouraging others to use open source is always (in my mind) a Good Thing.
UNIX Administratosphere 
There are additional clones of Matlab like, e.g.
Freemat, SciLab and JMathLib