Re: On the use of multiple PA prefixes or a single PI prefix for IPv6 multihoming

On 23-okt-04, at 8:00, <[email protected]> wrote:

>> The other reduction is that I'm not sure we need to solve
>> the "best path" problem but solving even the "a working
>> path" problem might be sufficient. That's at least my
>> preference.

> Exactly; my thought has always been that we need to avoid
> the non-working path(s) & that would be sufficient.

Avoiding non-working paths goes to the core of what's multihoming all 
about, so this is something we absolutely need. But I'm not sure it's 
all we need. What I'm worried about is the situation where there are 
multiple working paths, but there is a large difference in quality 
between them. For instance, one path has a capacity of 100 Mbps and 
another a capacity of 1 Mbps. In this case, selecting the latter path 
as part of regular operation would hardly be acceptable.

We _may_ be able to bring this problem back to manageable proportions 
by taking advantage of the information both ends have. So if either end 
knows that the second path is much slower, they'll avoid it as long as 
the first path is operational. This leaves only the cases where there 
is congestion in the core, which is pretty rare these days. (But this 
hasn't always been the case...)

Note that the whole reachability detection problem isn't as simple as 
it may seem at first, especially when we want to make use of 
unidirectional paths. Until not very long ago, I was favoring a 
"cartesian ping bomb" approach where there are reachability probes for 
all { src, dst } combinations. However, this can cause massive 
congestion when a link fails for a busy site, and it's also complex and 
most likely slow.

My current homework for the design team is to look how protocols such 
as OSPF and especially STUN implement similar features and see what we 
can borrow there. Unless this provides significant new insights I want 
so see if we can make a "be optimistic but check first" approach work. 
This means that when the reachability heuristics determine that there 
may be a path failure, we check the current path. If there is a 
failure, we check the next best path (for whatever "best" means 
locally) and so on, until we find one that works. In order to detect 
unidirectional reachability without excessive numbers of packets, each 
probe contains information about the last few multihoming signaling 
packets received from the other side. So unidirectional reachability 
won't be detected as fast as bidirectional reachability, but eventually 
it will.

I think we can probably tune this process such that significantly 
better (= lower RTT) paths are often discovered without any additional 
effort. This would be accomplished by having an artificially low 
timeout for the initial reachability test. (Successive ones should 
probably use an (exponential?) backoff.) So we send probe A->X and then 
25 ms later B->Y. If the difference between RTT is bigger than 25 ms, 
the second probe will be answered first and we're in business. But if 
the RTT difference is smaller than 25 ms we'll still get a reply for 
the second probe after the reply for the first probe was received, as 
long as this one took longer than 25 ms. This works pretty well for a 
certain range of RTTs, only with very short ones we don't see better 
paths and with very long ones we send too many packets. If we can make 
the initial timeout a function of a previously established RTT we can 
probably get around this.

Another way to get around selecting a working, but non-optimal path is 
allowing applications to set an "impatient" flag which then triggers 
more extensive path evaluation.  :-)