Or, like me, you only have one *transit* provider, but a number of (small) private peering connections. Yes, the use of private/fake AS space is feasible, but the same issues arise as with the misuse of RFC1918 space between organisaions. Peter ----- Original Message ----- From: "Wolfgang Tremmel, WT5-RIPE" <wtremmel@vianetworks.com> To: "Vladimir A. Jakovenko" <vovik@lucky.net>; "leo vegoda" <leo@ripe.net> Cc: <lir-wg@ripe.net> Sent: Tuesday, July 09, 2002 1:10 PM Subject: RE: [lir-wg] AS Number Policy
I'd say yes, also if it perhaps once was intended differently.
There are a number of ISPs out there which have for cost reasons only one upstream provider, but want to (also for cost reasons) peer at an exchange. This might not be the most redundant approach, but we should not tell them how to run their business.
What would happen if we'd not allow this? Just more wrong entries in the RIPE database.....
best regards, Wolfgang
-----Original Message----- From: owner-lir-wg@ripe.net [mailto:owner-lir-wg@ripe.net]On Behalf Of Vladimir A. Jakovenko Sent: Tuesday, July 09, 2002 1:08 PM To: leo vegoda Cc: lir-wg@ripe.net Subject: Re: [lir-wg] AS Number Policy
On Tue, Jul 09, 2002 at 11:53:14AM +0200, leo vegoda wrote:
Dear Colleagues,
Within the RIPE community the AS policy (ripe-245) is based upon RFC 1930. The RIPE policy states that "a network [is required] to be multihomed for an AS number to be assigned". However, there is no clear policy on returning unused AS numbers to the pool.
Would it be possible to define term multihomed more clearly? Does it mean that AS just should have more than one eBGP peer?
Lets see the following example:
+-----------+ +-------+ | AS-UPLINK | | AS-IX | +---------o-+ +o------+ | | +o-----------o+ | AS-CUSTOMER | +-------------+
AS-CUSTOMER is world-visible through AS-UPLINK and limited visible to some ASes through AS-IX.
Should AS-CUSTOMER be considered as multihomed?
-- Regards, Vladimir.