PI vs PA Address Space

peter at swan.lanl.gov peter at swan.lanl.gov
Fri May 19 22:07:55 UTC 1995

>>> As an aside, is anyone else besides Sprint behind this /18
>>> model?

The hard core /18 model ("we won't accept prefixes greater than length
18") is untenable and throws out one of CIDR's features.  It does not
allow for a time period where a customer is migrating from provider A
to provider B and will have end systems living within both provider
based prefixes at any instant during the migration.  

The user community should not be forced into flash cuts, and the
providers can make the needed overlap period of time work for bounded
time frames.  

At a minimum, the model needs to be /18+E (E==entropy due to customer


More information about the NANOG mailing list