Re: [5gangip] New Version Notification for draft-xyx-5gip-ps-01.txt

[Dino Farinacci <[e-mail]>]

>> Hi Tom,
>> 
>> An eNB today is strictly an L2 device today.
>> 
>> Assuming we change that, it may be more fit to say the eNB has a /48 and
>> assigns /64 to a UE per pdp / radio access bearer , that would align more
>> closely with today's architecture, and assuming each UE has exclusive access
>> to the /64, then rfc7238 may be relevant, as it is today.
>> 
> Thanks. So in that case if we let the locator becomes 48 bits which
> means we can have 65K UEs in the network with 80 bit identifier and
> assuming /64 bit assignment. That seems too little UEs for an ID/loc
> split addressing.

This is why an encapsulation data-plane should be considered, even with additional packet overhead, you get flexibility with the 128-bit EID and 128-bit Locator so you don’t have to make the tradeoffs above. 

And you’ll see other benefits at the same time.

Dino