Re: [lisp] LISP EID Block Size

[Rene Bartsch <[e-mail]>]

Am 2013-11-01 13:45, schrieb Sander Steffann:
> Hi,
> 
>> I want to ask everyone on the list: Which facts prevent a scaling 
>> experiment with the aim of global production state? In my opinion a 
>> /16-EID-prefix is perfect for that goal.
> 
> The problem is in that what you describe depends on public PITRs, and
> we have seen how badly that worked for 6to4 public relays. Running a
> public relay costs money (equipment, maintenance, bandwidth), and when
> nobody pays for them then we cannot expect any decent quality. And
> LISP will be blamed and seen as an unreliable protocol, just like
> 6to4. Relying on public relays is a very bad idea.
> 
> Now, if some big tier-1 transit networks start running production
> quality PxTRs (because PxTRs attract traffic, and their customers pay
> for traffic) then I can see some possibilities. If the LISP traffic
> volume increases then other networks might also start running PxTRs so
> they don't have to pay their transits for it, and then we are getting
> somewhere. But as long as 'public PxTR' means 'someone with good
> intentions but no real responsibility' then this will be a dangerous
> experiment for LISP...

That's an important argument.
We shouldn't rule out public PITRs because of the huge traffic to be 
expected, provide a /16 EID-block and hope it will attract operators of 
backbones and internet exchanges. Maybe we can define some Quality of 
Service rules for PITRs to discourage fun installations with low 
quality.

Best regards,

Renne