[nylug-talk] Does anyone have a case why we should still compile custom kernels

[Michael K Dolan]

> > for what type of use scenario and what distribution? In general, most
> > people have no need to compile a custom kernel these days if they're
using
> > a modern RHEL/SLES, but it depends on the intended usage...
>
> Is this a recent development? I thought that for >90%* of the users,
> and probably 98%* of business and consumer systems, a generic kernel
> will do everything, and has for about the last 6 years.
>
> -Peter
>
> * My percentages are completely bogus. I can't back them up with
>   anything and I won't try. It really just means "lots".

Actually, my guess is the percentage of business systems that work best
with a generic kernel is probably >99%. I didn't intend to convey that
there were many cases where you would want to compile a custom kernel - in
fact I mean the opposite - I see absolutely no reason to compile a custom
RHEL kernel unless you're doing something way out of the ordinary (and even
then there are only a few cases where that is required, many cluster users
these days don't even compile their own kernel)... The RHEL/SLES 2.6.x
kernels these days are fantastic - and they're very well tested.