Author: Robin 'Roblimo' Miller
Beyond that, I download and install a new piece of software once in a while, using the MEPIS-customized version of KPackage as a GUI front end to apt-get. Each installation takes me three mouse clicks, plus typing in my root password. My only other “customization” was setting up my bookmarks, email filters and “junk” controls in Mozilla, which I did about two years ago on one machine. Since then, all I’ve done is copied those files to each new computer I’ve tested or used for daily work.
Otherwise, my computers are as stock as stock can be, and they start and run as reliably as a diesel Mercedes. I spend no more time fiddling with them than I spend fiddling with my electric drill or circular saw. My computers and other power tools exist to serve me, not the other way around. I pick them up and use them when I need them, then put them away until the next time I have a task for which they are required.
Optimization as a hobby — or business
My personal computing needs are simple; I’m a writer and editor, not a sysadmin in charge of a 400-node supercomputing cluster, nor am I so in love with my computers that I want to treat them as the data processing equivalent of a hot-rodder’s pride and joy.
I have friends who spend endless hours extracting the last bit of performance from their racing cars — and their computers. They tune, tweak, modify, and test far more than they either drive or compute. Naturally, because of the endlessly mutable nature of GNU/Linux and open source software in general, many of these people are attracted to the open source world more than they are attracted to proprietary software, so Linux User Group email lists are full of messages back and forth from these people about how they managed to get a one percent performance increase by changing some obscure bit of code in a file the rest of us will never need to open.
This is probably true for most commercial GNU/Linux users, too. If the Web servers work reliably, why mess with them all the time? Sure, you need to keep up with security updates, just as you need to change the oil in your car regularly, but that’s no big deal if you use a well-established distribution. If server needs increase, buy another box. Or buy two. It’s usually cheaper to buy and set up more servers than to pay a skilled sysadmin a week’s salary to squeeze a little more performance out of existing hardware.
Naturally, there are circumstances — like supercomputing clusters and other large-scale installations — where performance tweaking is a better investment than simply throwing more hardware at a problem. And in these situations you want to have the most tunable, modifiable, tweakable software you can get — which means your operating system and applications software must be open source if you are going to get maximum performance from your system.
“You can” is not the same as “You should”
The hackneyed programming maxim says, “Just because you can do something doesn’t mean you should.”
In the case of GNU/Linux tweaking, you might say, “Just because you can modify your system doesn’t mean you must.”
To anyone who worries that if they switch to GNU/Linux they’re going to have to spend all their time messing with their system, I say, “If all you knew about cars came from watching NASCAR, you wouldn’t think about buying a car for everyday use.”
So don’t worry if it seems like half the Linux users in the world seem to have more in common with car performance freaks than with normal people. You don’t need to do all that to run Linux either at work or at home. For most of us, a stock distribution is fine, even if the tweakers sneer at us the same way someone who owns a tricked-out Lexus sneers at a perfectly sensible, one hundred percent stock Honda Civic.