Policies: Routing a subset of another ISP's address block
We have a situation where we have a client who wants to be dual-homed for redundancy. They are not large enough to get addresses from ARIN. Given that they are wanting us to allow another provider to route a subset of one of our address blocks(5 /24's out of a /16). Looking for some recommendation/dangers and general policies in reference to this. Thanks for any input. If this is the incorrect list to post this on please let me know. thks -- David Harrison
On Wed, Apr 05, 2000 at 08:33:25AM -0400, David Harrison wrote:
We have a situation where we have a client who wants to be dual-homed for redundancy. They are not large enough to get addresses from ARIN. Given that they are wanting us to allow another provider to route a subset of one of our address blocks(5 /24's out of a /16). Looking for some recommendation/dangers and general policies in reference to this. Thanks for any input. If this is the incorrect list to post this on please let me know.
Refuse to do it, the customer must get PI addresses for this purpose. /Jesper -- Jesper Skriver, jesper(at)skriver(dot)dk - CCIE #5456 Work: Network manager @ AS3292 (Tele Danmark DataNetworks) Private: Geek @ AS2109 (A much smaller network ;-) One Unix to rule them all, One Resolver to find them, One IP to bring them all and in the zone to bind them.
That's simply not true. Many ISPs will advertise another's netblocks for a mutual downstream. The client doesn't have enough IP space to qualify for PI space in any case unless they utilize a /21 to 80% while being multihomed. I don't see the logic behind refusing the customer a request of this sort. Daniel Golding Senior Network Engineer NetRail, Inc. On Wed, 5 Apr 2000, Jesper Skriver wrote:
On Wed, Apr 05, 2000 at 08:33:25AM -0400, David Harrison wrote:
We have a situation where we have a client who wants to be dual-homed for redundancy. They are not large enough to get addresses from ARIN. Given that they are wanting us to allow another provider to route a subset of one of our address blocks(5 /24's out of a /16). Looking for some recommendation/dangers and general policies in reference to this. Thanks for any input. If this is the incorrect list to post this on please let me know.
Refuse to do it, the customer must get PI addresses for this purpose.
/Jesper
-- Jesper Skriver, jesper(at)skriver(dot)dk - CCIE #5456 Work: Network manager @ AS3292 (Tele Danmark DataNetworks) Private: Geek @ AS2109 (A much smaller network ;-)
One Unix to rule them all, One Resolver to find them, One IP to bring them all and in the zone to bind them.
We usially allowed to do it after notifying us (announce the specifics from our blocks) but without any responcibility for the reults. It's strange but in real world this method works very well. ----- Original Message ----- From: "Jesper Skriver" <jesper@skriver.dk> To: "David Harrison" <david.harrison@interpath.net> Cc: <nanog@merit.edu> Sent: Wednesday, April 05, 2000 6:01 AM Subject: Re: Policies: Routing a subset of another ISP's address block
On Wed, Apr 05, 2000 at 08:33:25AM -0400, David Harrison wrote:
We have a situation where we have a client who wants to be dual-homed for redundancy. They are not large enough to get addresses from ARIN. Given that they are wanting us to allow another provider to route a subset of one of our address blocks(5 /24's out of a /16). Looking for some recommendation/dangers and general policies in reference to this. Thanks for any input. If this is the incorrect list to post this on please let me know.
Refuse to do it, the customer must get PI addresses for this purpose.
/Jesper
-- Jesper Skriver, jesper(at)skriver(dot)dk - CCIE #5456 Work: Network manager @ AS3292 (Tele Danmark DataNetworks) Private: Geek @ AS2109 (A much smaller network ;-)
One Unix to rule them all, One Resolver to find them, One IP to bring them all and in the zone to bind them.
Well from sentiment I pick up from nanog lists and last nanog meeting it does not seem many provides are actually worried about the size of the internet routing table anymore. Use to be the main objection to routing table growth was the fear of core routers become expensive space heaters . I am inferring here that routers have caught up and then some handling larger and larger tables. So why is there still so much resistances to supporting multi-homed customers that, shock horror, involves providers advertising more discreet routes that are in the middle of their cider blocks? I am guessing administrative overhead is main objection now. The whole micro-allocation conversation show provider willingness to allow growth in the routing tables. I have my flame retardant suit on so go for it. The next good question is what do providers think the remaining headroom, prefix and route wise, is remaining in the main stream platforms (rsp/4, gsr, m40,etc). I am assuming this is not main mem size issue since routers these day can take in obnoxious amounts of memory. Guessing again more of a cpu capacity and (for dCef type implementation) line card memory related issue. - Dustin - ps. Excuses like my network still runs on an AGS+ will not cut it. :-)
On Thu, Apr 06, 2000, Dustin Goodwin wrote:
Well from sentiment I pick up from nanog lists and last nanog meeting it does not seem many provides are actually worried about the size of the internet routing table anymore. Use to be the main objection to routing table growth was the fear of core routers become expensive space heaters . I am inferring here that routers have caught up and then some handling larger and larger tables. So why is there still so much resistances to supporting multi-homed customers that, shock horror, involves providers advertising more discreet routes that are in the middle of their cider blocks? I am guessing administrative overhead is main objection now. The whole micro-allocation conversation show provider willingness to allow growth in the routing tables. I have my flame retardant suit on so go for it.
Just because people might not fear their routers melting under large network tables doesn't mean tomorrow they want 100,000 /30's in their routing tables. The resistance is there so people who really positively have no other choice - there are lots of other possibilities, and if some unexpected sideeffect of lots of /30's pop up, the entire internet suffers rather than just one customer. Adrian
participants (6)
-
Adrian Chadd -
Alexei Roudnev -
Daniel L. Golding -
David Harrison -
Dustin Goodwin -
Jesper Skriver