[cgl_discussion] Booting Linux faster

Patrick Mochel mochel at osdl.org
Fri Sep 19 10:31:17 PDT 2003


> > http://midori.transmeta.com/pub/midori-1.0.0-beta3/apps/quickinit_ML1.0.0-beta3-2.mlz
> > 
> > It's just a tar file, and is small enough to be self-explanatory. If 
> > anyone is interested, feel free to email for more information about
> > it. 
> 
> At least to me the problem here is not lack of interest (well, time
> maybe but that is another issue), but rather the political ramifications
> with LSB.

I didn't realize there was a problem with the LSB to begin with. :) 

Who cares what the political ramifications are? The problem at hand is 
that Linux could boot faster. There are many people who request such a 
feature from embedded folks to carrier grade folks to yours truly. I hate 
waiting the 23 seconds it takes from the time LILO loads the kernel to the 
shell prompt I get. 

This can be solved by parallelizing the init scripts, as has been proven
with three separate implementations of an init replacement. It has not
been widely adopted because, AFAIK, no one has created a drop-in solution
that runs transparently on modern distributions, and is backward
compatible with the current setup. 

That is the challenge. The LSB isn't even a factor. If someone were to
step up and polish something well enough, poeple would listen and pick it
up and start using it. It would be widely accepted and all the 
non-technical issues would resolve themselves. 


	Pat




More information about the cgl_discussion mailing list