[rrg] Fundamental objections to a host-based scalable routing solution

[William Herrin]

On Tue, Nov 25, 2008 at 12:34 PM, Luigi Iannone
<luigi at net.t-labs.tu-berlin.de> wrote:
> William Herrin wrote:
>> Who said anything about sending extra packets? The payload packets are
>> the test. The acks (or their absence) are the response to the test.
>
>
> This if you assume symmetric traffic (e.g., TCP) and symmetric routing
> (i.e.,  A -> B = B -> A).
> IMHO you can't make this assumption.

Luigi,

One of us is very confused. I was talking about host-level traffic
where exactly two hosts using multiple locators are talking to each
other. What are you talking about?


On Tue, Nov 25, 2008 at 12:39 PM, Dino Farinacci <dino at cisco.com> wrote:
> And assumes that the data rate of the host returning traffic on the
> symmetric path is within your switchover budget. And what if traffic is
> unidirectional?

Dino,

It couldn't be purely unidirectional or you wouldn't have been able to
look up the GUID to find the locator in the first place.

That is one of the constraints on a host-based solution. A trivial
host can't do unicast fire-and-forget the way, for example, SNMP traps
do. Multicast and anycast fire-and-forget is still possible.

Regards,
Bill Herrin



-- 
William D. Herrin ................ herrin at dirtside.com  bill at herrin.us
3005 Crane Dr. ...................... Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004