Re: [savi] Binding establishment based on data or control packets?

[Christian Vogt <christian.vogt at ericsson.com>]

Guang Yao wrote:

> If the SAVI device asks the node before deleting bindings, it is  
> safe in
> most cases and only fails in some specified examples. If the SAVI  
> device
> doesn't ask the node and delete bindings unilaterally, it may fail  
> with
> great possibility. So I don't think this is a good example to prove
> querying the node is useless.

I think the conclusion regarding the use of probing prior to removing
soft-state bindings should be that (1) SAVI devices do probe to minimize
premature binding removal, but that (2) we cannot rely on a binding to
exist if a host is using the bound address.  In other words, the probing
can only be an optimization.

Does this make sense?

- Christian