Brian, > "Call home" is IMHO a fairly radical departure for SNMP and > raises trust model questions that I don't find easy to get > hold of. It seems quite distinct from both firewall traversal > and NAT traversal, conceptually, even if they might be > a side-effect of calling home. I do not believe - nor have I seen document demonstrating that - you can practicably achieve the two classic functional goals of query/response and unsolicited notification without implementing CH in the face of a firewall/NAT. Another way of putting it is that if firewalls are to be dealt with (and in this day and age we should presume this), this functionality has been prematurely ruled out of scope and must be considered in order to that need ANYWAY. Eliot _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf