14 Jan
2016
14 Jan
'16
10:32 p.m.
On Jan 14, 2016, at 4:11 PM, Johnny Billquist <bqt
at softjar.se> wrote:
On 2016-01-14 21:50, Peter Lothberg wrote:
Could you explain why you think this is better?
Now, I just did realize one possible reason, and that is since all bridges use Update as
a hub, the UDP packets run through there, even if talking between two nodes in the US.
That is suboptimal. Now, the bridge don't have to be setup that way, it's just
been the case that people have decided to use Update as the hub. It would make much more
sense to have a couple of bridge hubs in the US - say one west, and one east. Have west
bridge connect to east bridge, and have east bridge connect to Update. And then people
connect to the hub in reasonable vicinity.
But the bridge as such, are not worse than the multinet links. If your only reason is
because of the current topology of the bridge itself, then yes, for some it will be a win
to favor the multinet links. Mostly people in the US. For others, it might definitely be
worse.
It sounds like you're reinventing routing... but if you want routing, why not have the
routing layer do it?
The bridge is a nice mechanism, and a very simple way to do straightforward things. But
if you run into situations where it isn't the right tool, having the DECnet routing
mechanisms solve it for you is probably the simplest answer.
paul
As for
cost tweaking for other areas (areas without multiple area
routers that is), I do not believe that tweaking the costs to favor
Multinet links will improve life. But I'm interested to hear theories
that claims otherwise, and we can try figuring out if they make sense.
If nodes that are both on the bridge-ethernet and also acts as
multinet hubs had a resonable high metric on their ethernets...