[rrg] Remote ACLs [Proposals which match rrg architectures.htmlpls check the page]

[David Conrad]

Dino,

On Jan 4, 2009, at 5:19 PM, Dino Farinacci wrote:
>> If we want to use ID in the ACL, should it also have some structure  
>> in
>> semantics? since the flat ID will cause a scaling issue with ACL  
>> and a huge
>> burden for manually configuration.
> If EID-prefixes are allocated from RIRs, we can have structure.

Depends on the allocation rules for EIDs, which obviously don't exist  
now.  Right now, the RIRs try to juggle conservation of address space  
with conservation of 'routing slots' with a bias towards the latter,  
resulting in allocations of power-of-2 size blocks aligned on CIDR  
boundaries.  If the RIRs were to allocate EIDs, they would not need to  
be concerned about conservation of 'routing slots', hence non-CIDR  
aligned and non-power-of-2 blocks can be allocated.  If this isn't  
what people want, the RIR public policy process would have to be  
engaged to come up with the EID allocation rules.

Regards,
-drc