[ale] nfs & collisions

[Zachary Link]

Putting the two machines on a single hub AND disconnecting other hubs
could dramatically improve performance if the other hosts on the other
hubs were using a decent amount of bandwidth.  I think that was what the
suggestion was.

And the short answer to your question about the hub and local addresses is
there is no local/remote distinction in a hub (or switch for that matter).
 A hub broadcasts every packet out every port, except the one it came in
on.  A switch keeps a table tracking which port every mac addrress is on. 
In the case of a host on a "remote" switch, the uplink port will just have
a lot more mac addresses associated with it than one with a single server
on it.


Zack



<quote who="Jason Day">
> On Mon, Feb 24, 2003 at 05:10:53PM +0000, hbbs at attbi.com wrote:
>> If you're trying to push ~20GB between two machines through four
>> unswitched  (and I presume 10base-T) hubs, then, yeah, it's gonna
>> collide like a SOB and  take 31 hours.  At the very least, get the two
>> machines on the same hub, but,  better yet, put a 10/100 switch in
>> place.  Just because the four cascaded hubs  works doesn't mean it
>> works efficiently.  The latency is just going to be awful.
>
> <tangent>
> Would putting the two machines on the same hub really make any
> difference?  I thought unswitched hubs always broadcast all traffic over
> all ports.  Or is an unswitched hub still smart enough to broadcast the
> traffic only on the local ports if the destination address is local?
> Just curious.
> </tangent>
>
> Jason
> --
> Jason Day                                       jasonday at
> http://jasonday.home.att.net                    worldnet dot att dot net
>
> "Of course I'm paranoid, everyone is trying to kill me."
>     -- Weyoun-6, Star Trek: Deep Space 9
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://www.ale.org/mailman/listinfo/ale



_______________________________________________
Ale mailing list
Ale at ale.org
http://www.ale.org/mailman/listinfo/ale