DragonFly commits List (threaded) for 2007-07
Re: cvs commit: src/sys/kern kern_fork.c kern_usched.c src/sys/sys usched.h
On 7/2/07, Hasso Tepper <firstname.lastname@example.org> wrote:
Matthew Dillon wrote:
> Presumably if one really had 31 cpus he would also have sufficient
> memory and not the measily 64M-1G that we specify when we run the
> vkernel. Plus I don't think a 31-cpu machine would be used to route a
> full BGP table. When Jeff and I discussed this last year we figured
> that it was a moot issue.
At first "full BGP table" isn't the largest one in the world. I have seen
routers which have more than million routes in the table, providers can
do funny things in their AS's ;).
Heh. I love those route-distinguishers :-)
And about many-many routes in 31-cpu machine ... depends what benefits
would be. I can easily think^Wdream about configs where I'd love to. But
yes, I don't see much benefit from all of them having _one_ large routing
table. Much more I'd love to see them having different tables - thinking
about virtual routers (and multiple routing tables in them).
I share your vision about virtual/logical routers.