[HECnet] Multinet peerings...?

[Steve Davidson]

This past Sunday, Johnny and I spent a fair amount of time looking at 
circuit costs (and performance) between SG1:: and JOCKE:: while testing 
his new Multinet-tunneling code.  The findings were a bit strange to say 
the least.  At the moment, the LAN circuit costs here are set to "4" to 
reflect what the VMS machines set their network interfaces to.  It seems 
that RSX sets costs to "3".  Why are they different?  Who knows!

The Multinet circuit costs coming out of here are "4" or less depending 
on various factors.  They include but are not limited to: circuit 
speeds, number of links to a leaf node (site), and types of links.  I 
treat this site as a hub because it services so many leaf nodes 
(sites).  I treat all other sites (except LEGATO:: and MIM::) as leaf 
nodes (sites).  LEGATO:: and MIM:: are treated as hubs (or peers).  In a 
2 weeks I will be back from vacation and will be prepared to look 
further into possible circuit cost changes.

Bandwidth at this site (either coming or going) is no longer a problem.

Stay tuned...  :-)

-Steve

On 1/13/2016 20:52, Peter Lothberg wrote:
> 192.108.200.211  -- 130.238.19.212	Decnet COST 30
>