Re: ISP failures and site multihoming [Re: Enforcing unreachabilityof site local addresses]

[Iljitsch van Beijnum <iljitsch@muada.com>]

On 20 Feb 2003, marcelo bagnulo wrote:

> > If each host is connected over two
> > service providers there are four possible paths the hosts can switch
> > between on a per-packet basis. Then the only problem becomes detecting
> > failure. The end hosts are in an excellent position to do this without
> > having to generate keepalive messages;

> How?

With TCP you simply monitor incoming ACKs. If there aren't any for a
while, when they are expected, something must have gone wrong -> switch
to an alternate path. UDP-based protocols can use a similar mechanism,
but they often already know how to fall back to a secondary address.

The main problem is not switching to an alternate path too aggressively
when some packets are lost or there is an increase in delay.