Re: [RAM] The mapping problem: rendezvous points?

[David Conrad <drc at virtualized.org>]

Dave,

On May 8, 2007, at 1:27 PM, Dave Thaler wrote:
> Applications generally work fine with random loss.
> They work less well with _deterministic_ loss.

I would've thought the opposite would be true.

> Some common application models that are adversely affected by PULL:
>
> 1) Client resolves a name to a list of addresses, and tries each one
> (with some timeout since there's no guarantee of response) until a
> connection succeeds.
> ...
> 2) Client resolves a name to a list of addresses, and tries each
> one in parallel and chooses the first one to succeed.
> ...

Since both of these are already relying on a pull mechanism at the  
initiation of a transaction, it isn't clear to me how they would be  
significantly more adversely affected by using a pull-based mapping  
mechanism.

> 3) Server responds to a simple client request via an asymmetric path.
> No DNS occurs here, and the response goes via a router not involved
> in the client-to-server direction.

This case would indeed cause an increase in latency.

> The above are just examples of common application classes which show
> why I personally consider PULL schemes to be a non-starter.

I was wondering if you had concrete examples of existing applications  
that would fail with an increase in initial packet latency.

Rgds,
-drc


_______________________________________________
RAM mailing list
RAM at iab.org
https://www1.ietf.org/mailman/listinfo/ram