[e2e] QoS vs Bandwidth Overprovisioning

David G. Andersen dga at lcs.mit.edu
Sun Apr 29 13:27:40 PDT 2001

But not all, at least from the perspective of an overlay network
on the Internet..  In some empirical measurements for the RON project,
we've found that there are often trade-offs between the lowest-latency
indirect route (e.g. "route through MIT to get to Utah") and the
lowest-loss indirect route.

That may be a feature of the multiple link aggregation we see by operating
as an overlay network, of course, and not actually present on a 
per-physical-link basis.  Especially since "Highest physical
bandwidth" link and "lowest loss link" are not necessarily


RJ Atkinson just mooed:
> >
> >we went through quite a bit of effort on that a few years back, and finally pulled them out of OSPF because the least monetary cost route, the highest bandwidth route, the lowest delay route, and the most reliable route were all the same route.
>         This was not true for the DDN trans-Pacific network,
> which had deployed OSPF with ToS bit support.  That deployment
> particularly used the delay metric to help select between
> satellite hop and undersea cable.  The main problem that I
> saw with that part of OSPF was that cisco flat refused to 
> implement that part of the spec (at that time), so if one 
> wanted to implement OSPF with ToS routing support, 
> one had to eliminate a major competitor from the bidding 
> process.  Sigh.
>         I do believe that in many networks, all three paths
> would be the same, btw.
> Ran
> rja at inet.org

More information about the end2end-interest mailing list