DSL Ideas and Suggestions :: It could become real!
I've discovered that my vision is closer to reality than I thought.
Remember, my vision is of a light weight Linux distribution (DSL, of course) with openMosix clustering added on. The first papers I read _only_ discussed clustering across an isolated Lan segment. However, I've found that openMosix can operate across the Internet, at the price of added security.
openMosix works by:
1. Migrating only the processor stack, registers and instruction pointer from node to node.
2. Transmitting memory pages over the wire _only_ when there is a page fault.
This is all done by patching the Linux kernel.
The cluster management tools could all be downloaded as *.dsl or *.tar.gz files.
Since my vision is of ordinary users _donating_ processor cycles, the only parts of the project left are:
a. Hard-wiring security settings so the end user can _not_ change them.
b. Figuring out an easy way to manage initial scheduling.
The end goal is a 200 Mb live CD (Shirt pocket size!) with a second pocket sized CD of clustering tools and a website where initial scheduling could be done.
N.B. ----------------------------------------------------------------------------------
If the patches were worked into the DSL build, I just might be able to do the rest myself!
----------------------------------------------------------------------------------------
The size will be beyond the 50 Mb limit of DSL, but in keeping with DSL-N.
We could call it Super Damn Small Linux (SDSL) and market it to computer science students, business students (to analyse the stock market), physics students, et cetera!
I'll keep you all posted on my progress...
(John and Roberts, I'll keep you especially posted.)
original here.