Linux.com

Kernel bloat.

Posted by: Anonymous Coward on July 26, 2004 08:58 PM
I'm glad to see someone addressing the problem. If the current trend toward bloat (and not just in the kernel code) continues, Linux will lose much of its value.

Lean and mean isn't just about running on a 486SX or inside some device as an embedded OS. It is also about wringing real performance out of any compatible hardware. It is about empowering the poor who cannot afford Win XP compatible desktops, Win 2003 servers, and who won't be able to afford Longhorn compatible hardware tomorrow. It is about preventing waste in a throw-away society and curbing the upgrade cycle and planned obsolescence.

I'd like to see a new dedication within the Linux community, and that is while all the great new enterprise features are cobbled on, and as new architectures are supported, that the community also rigorously support hardware built 10 years ago.

Perhaps the 386 processor has outlived its usefullness. And maybe it is time to move forward to supporting no less than the 486 and 8 MB RAM. But lets draw the line there for at least another 5 years before deciding the 586 is the oldest processor, and 16 MB RAM is the least supported.

This is a world of 6 billion souls and many are struggling to join the world economy. Amortized hardware running Linux around the globe will do much to establish Linux and Linux applications as the world standard, provided Linux remains useable on that old hardware.

Most of us, if we live and work in the industrialized world, overlook that. By falling into the trap of code bloat, Linux offers one less incentive to break the dependence on Windows as a standard.

Those who don't believe there is still a need for Linux to run on 486 -- in both desktop and server roles -- need to travel some of the less affluent corners of our planet.

#

Return to OLS Day 3: Failed experiments, Linux-Tiny, and the Linux Standard Base