Re: Is Hurd good?
Catalin <ady982@ploiesti.astral.ro> wrote:
> So the HURD will be the most stable and the most
secure operating system
> on earth but also the most slowest operating system
on earth (because
> the request travels a longer way)?
This just prooves that every-day science is broken.
Some performance critical applications (not servers)
would benefit from the Hurd's design (google for
"User-Level Networking" and check the performance
results). It just contradicts what you said, packets
would have a longer path from user-space applications
to the in-kernel stack (this involves expensive
copying) un-like in most ULN stacks. Intel, Microsoft,
Compaq (HP atm) and others from the VIARCH alliance
beleave that ULN is the future, check U-Net (I forgot
the URL, google it :p) as a good example of how fast
ULN could be, they managed to get a fast ethernet NIC
to compete with a 155MB ATM connection.
Usuauly design mistakes are the reason of degrading
applications/servers's performance. One good example
of this is Foxnet's ULN stack
(http://www-2.cs.cmu.edu/~rwh/papers/foxnet/hosc.pdf
) the guys actually cared about getting it to work in
any form, but I think that with some improvments
(namely, a better packet classifier to hand packets to
the upper-layers in a faster scheme) it would compete
with other in-kernel stacks.
So, I guess I'll just agree with wolfgang, and wont
blame the Hurd for being slow by design. After all,
everything has advantages and disadvantages. But after
all, I could be wrong and anyway, I have nothing to do
with OS design.
Cheers,
kotry
__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com
Reply to: