> ... due to the deleterious effect a lack of ability to aggregate > addresses and summarize routes have on both the size and CPU/Memory > requirements necessary to hold and calculate routing tables. After so many years of this I'd have thought someone would have had a go at tackling it, surely a hole in the market? > High-end routers also tend to be very expensive meaning that if > steps were not taken to minimize portable allocations, providers would be > forced to purchase ever increasingly expensive routers at ever shortening > intervals. I thought they were already, backbones have grown a lot in the last few years with >= OC3 becoming common. Certainly wasn't conceivable that I would have any a few years ago but do now. One thing that's never been clear is how many extra routes are we talking about? How many ISPs would advertise some if they could or if they had to prove they needed to (i.e. had more than one upstream), Is it 10000, 100000 or more? Looking at the CIDR report it appears 4000 or so entries could be freed just by better agregation, that could be 4000 happier ISPs instead for no increase. It also suggests 50% of ASes only announce one prefix already so that's lots of ISPs that have worked their way round this limitaion by getting enough space (one way or another). > This cost of course would necessarily be shifted to the consumer. That's not a reason for not doing it though, if the costs go up the customer gets to chose, 1) pay more and have better connectivity 2) pay less and not. It's not really free market otherwise is it? > RIPE for example Not a good argument, it seems we're relatively fortunate in having RIPE compared to the examples given for elsewhere. > I'd like to see the organizations with "grandfathered" large allocations > be asked to justify those allocations or have them taken back. If we we really ran out there'd be a use it or lose it test? Until then the two arguments are being mixed, there is enough space but it's being used in a manner some people don't like and routing table size is used to justify not changing that. regards, brandon
