[FoRK] DEC in the past; was Re: What do you think of my

Eugen Leitl eugen at leitl.org
Tue Jul 30 00:04:13 PDT 2013


On Mon, Jul 29, 2013 at 08:47:20PM -0700, Joseph S. Barrera III wrote:

> I would argue that (relatively) fast fork() was a seductive trap
> that took years to get away from. Modern server architectures use
> process pools or thread pools instead of relying on cheap process
> creation (which in general is very difficult in any OS).

Threads don't work on kilonode and meganode/giganode architectures, 
so this is a stillbirth. I keep telling our developers that they
shouldn't complain if their multithreaded stuff runs slower on
new servers (sea of small nodes on a mesh) than on legacy ones.


More information about the FoRK mailing list