Peter Lothberg <roll@bsd.stupi.se> writes:
2) networks having an existing AS-number in database but reflected differ ently in the routing tables (the can sometimes be a config error).
128.86.0.0 in AS786 (database) and AS1755 (BGP)
This is due to the fact that 128.86.0.0 is used as the London DMZ, and even if the EBS receives if with AS786 from the Janet RBS, it will prefer the cheapest route, locally connected to the EBS when announced to the outside world.
From ripe-81:
... o An IP network number can and must only belong to one AS. This is a direct consequence of the fact that at each point in the Internet there can be exactly one routing policy for traffic destined to each network. In the case of the IP network which is used in neighbor peering between two ASes, say at the border between two ASes, a conscious decision must be made as to which AS this IP network number actually resides in. ...
So either the JANET RBS or the EBS should stop injecting this net into EBONE. Any problems with that?
Technical problem, if we announce it with the Janet RBS, other boxes on that DMZ might be dependent on the RBS. Political problem, they might not want to change the network number of that network. Suggestion, use the Ebone network for the DMZ... -Peter
Peter Lothberg <roll@bsd.stupi.se> writes:
So either the JANET RBS or the EBS should stop injecting this net into EBONE. Any problems with that?
Technical problem, if we announce it with the Janet RBS, other boxes on that DMZ might be dependent on the RBS.
ACK that. So it should be injected from the EBS. As an aside: ---- When writing ripe-81 we did consider similar cases and came up with a generic one: DMZ1 DMZ2 AS A ------ AS B ------ AS C You have to decide into which ASes to put the DMZ networks. One thing that is worth considering here is that if ASA and ASC touch in one and the same router in ASC in traceroutes only the nets DMZ1 and DMZ2 are visible. If DMZ1 is in AS A and DMZ2 is in AS C the traceroute might show: ... DMZ1 (in AS A) DMZ2 (in AS C) ... And it is by no means obvious from this that traffic has crossed AS B.
Suggestion, use the Ebone network for the DMZ...
Looks right.
Peter Lothberg <roll@bsd.stupi.se> writes: * > * > > Peter Lothberg <roll@bsd.stupi.se> writes: * > > > 2) networks having an existing AS-number in database but reflected * differ * > > ently * > > > in the routing tables (the can sometimes be a config error). * > > > * > > > 128.86.0.0 in AS786 (database) and AS1755 (BGP) * > > * > > This is due to the fact that 128.86.0.0 is used as the London DMZ, an * d * > > even if the EBS receives if with AS786 from the Janet RBS, it will * > > prefer the cheapest route, locally connected to the EBS when announce * d * > > to the outside world. * > * > >From ripe-81: * > * > ... * > o An IP network number can and must only belong to one * > AS. This is a direct consequence of the fact that at * > each point in the Internet there can be exactly one * > routing policy for traffic destined to each network. In * > the case of the IP network which is used in neighbor * > peering between two ASes, say at the border between two * > ASes, a conscious decision must be made as to which AS * > this IP network number actually resides in. * > ... * > * > So either the JANET RBS or the EBS should stop injecting this * > net into EBONE. Any problems with that? * * Technical problem, if we announce it with the Janet RBS, other boxes * on that DMZ might be dependent on the RBS. * * Political problem, they might not want to change the network number of * that network. * * Suggestion, use the Ebone network for the DMZ... * * -Peter Okay, You are right we need to make note of these implications in the revision of ripe-81. In the 128.86.0.0 case this is not a problem politically or otherwise and the DMZ is an ebone net so not problem but the interconnect network is still an interesting one. However, there are only a few cases in the world where this is an issue. The GIX net is the most interesting of course. Basically, a decision has to be made as to who announces it. --Tony.
participants (3)
-
Daniel Karrenberg -
Peter Lothberg -
Tony Bates