On Fri, Jun 20, 2008 at 9:35 AM, Scott Brim <[EMAIL PROTECTED]> wrote: > On 6/19/08 10:44 PM, William Herrin allegedly wrote: >> >> The requirement is that end users (meaning folks who operate servers >> in this case) be able to change service providers: >> >> 1. Without a major overall effort, and >> 2. Without requiring any changes outside of the end user's >> administrative control. > > I don't believe the second one. First there's simple lower layer > connectivity -- of course you need permission from a NSP if you want to > receive traffic. Second there's routing. I see no way for packets to me to > traverse an intermediate provider without at least some node under that > provider's control being configured differently.
Scott, Allow me to clarify: the actions of an entity from which you purchase service are under your administrative control for the purposes of establishing criteria #2. You're paying them to do as you require. Arguably this continues upstream following the cash though as a practical matter you lose control more than a couple orgs away. Entities which purchase service from you are not under your administrative control. Entities with whom you have no fiduciary relationship whatsoever are not under your administrative control. With that clarified definition of "without requiring any changes outside of the end user's administrative control," do you have any further objections to this requirement? Would you like to rewrite the statement so that its more clear that one's vendors fall within one's administrative control? BTW, my counter proof to Tony's claim only addressed single-homed cases where PI is a necessary consequence of the requirements. Nearly all multi-vendor multi-homed cases require PI to function to an appropriate standard as well. Brian: This argument you see now is a fine example of why it's appropriate to escalate "PI for servers" to architectural requirement status. More than a few very smart engineers want NAT to have solved the need for PI but operational experience in the decade plus since NAT's invention very clearly show us that it has only solved the need for PI in a limited set of scenarios. Movement on the PI requirement since NAT's full deployment has been largely retrograde: changing the IP address of a high-volume mail server has become astonishingly hard because of all the spam filtering, as is corralling the 500 customers all of whom have forgotten the Godaddy passwords they need to update the A records for their hosted web service since they do that with Godaddy now instead of hosting their DNS with you. We're well past the point where "PI for servers" should be considered an architectural requirement until proven otherwise. Regards, Bill Herrin -- William D. Herrin ................ [EMAIL PROTECTED] [EMAIL PROTECTED] 3005 Crane Dr. ...................... Web: <http://bill.herrin.us/> Falls Church, VA 22042-3004 -- to unsubscribe send a message to [EMAIL PROTECTED] with the word 'unsubscribe' in a single line as the message text body. archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg
