Re: [WG-BB] Draft documents from Tokyo meeting

[MAEMURA Akinori <maem@maem.org>]

apology for this late reply.

In message <4.3.2-J.20010627102127.05d68100@localhost>
   "Re: [WG-BB] Draft documents from Tokyo meeting"
   "Takashi Arano <arano@byd.ocn.ad.jp>" wrote:

| At 01:08 01/06/27, Yoshihiro Obata wrote:
| >At 00:37 01/06/27 +0900, Takashi Arano wrote:
| >
| >> >>Furthermore, in such a sense, the first allocation rule
| >> >>can be applied also when allocating for new PoP.
| >> >>Do you think if this is true?
| >> >
| >> >PoP is a part of a network and normally it is not a service.
| >>
| >>Of course. What I try to say in addition to your comments
| >>is that when an existing ADSL ISP opens a new PoP,
| >>the ISP can assign /24 per CMTS for this PoP.
| >
| >   The size of wholesale and a PoP is much different.  I am not 
  > sure whether we need to apply this rule to justify expansion of 
  > service areas. If we do so, we need some measurement and limit for 
  > the size.
| 
| OK. If other people agree with Obata-san, I will withdraw my additional
| comments.
| 
| BTW, is it OK to incorporate Obata-san's idea to the proposal?
| 
I think it's OKay.  Since the usage is completely different 
>from their existing services, they must the same reason that
they cannot estimate the growth as an emerging ADSL based
ISP would have.



| 
| >> >  When an ISP uses a wholesale ADSL provider, they need to allocate 
     > a large block of IP addresses to a completely independent network, 
     > which normally has a different allocation policy due to its nature 
     > of permanent connectivity.
| >>
| >>I do not understand what is difference in cases of using a wholesale
| >>ADSL provider.
| >
| >   In our case or NTT Flets service, a large ISP is expanding several 
  > thousand subscribers per month, which might increase a few more times 
  > by the end of this year.  Nobody can estimate clearly such growth and 
  > what is worse is that it normally comes out from one PoP in Japan 
  > since they are combined with a backbone, which makes it more difficult 
  > to justify compared to dial up ISPs where you can distribute the 
  > network design among several PoPs.
| 
| I still don't understand. Our proposal assigns fixed size in the first time
| just because of the reason Obata-san mentioned: difficulty of estimation.
| 
As Obata-san mentions actually this kind of estimation is 
very difficult.  I think and believe the other people at 
that workshop also think that the excellence and essence of 
this proposal is to allocate/assign a fixed size block 
without any justification and to observe the usage after 
allocation/assignment.  This can reduce a big burden of
both LIR and NIR/RIR keeping the assignment appropriate.

I believe Obata-san will like this.


Best Regards,
-----
MAEMURA Akinori                                maem@maem.org
-
- This list (wg-bb) is handled by majordomo@lists.apnic.net