On Fri, 2006-09-15 at 21:01 +0200, Gert Doering wrote:
What I am really worried about is people getting "lots and lots" of PI, and using multiple routing table slots, instead of getting a reasonable chunk of addresses (however named), and announcing only *one* route.
This is a critical concern. As it stands, 2006-05 will do little but encourage rampant use of PI space by removing the psychological barrier of lying to RIPE. This is a bad thing: routing tables have once again grown to the extent that they have burst commonly used equipment - the current problem is with dual-stacked SUP720 pfc3a/b modules which now do not have enough tcam to deal with > 192K ipv4 prefixes in their default configuration (although this can be manually tweaked at the inconvenience of a reboot). But we all know this already. I've previously expressed opinions on address-policy-wg about the necessity of charging for PI space, both as a means for discouraging its assignment and for providing a legitimate means to reclaim dead IP space. I have to be honest here and say that this issue needs to be addressed rather more urgently than the proposals in 2006-05. We can continue to lie to RIPE on PI application forms, and they will continue to pretend to believe us so long as the figures add up correctly. But in the interim, PI address space is being lost from the global v4 address pool like a bad memory leak which everyone knows about but no-one wants to fix. Nick