Re: [Casm] I-D Action: draft-li-casm-address-pool-management-arch-00.txt

[Brian E Carpenter <[e-mail]>]

Hi,

Thank you for this draft. I have a few general comments, let me start with the Abstract:

>    it is complicated to
>    manually configure the address pools on lots of Broadband Network
>    Gateways(BNGs) for operators.

Not only for BNGs. I believe the same problem arises in any kind of large
network. Also, although the IPv4 exhaustion is a problem, so is the enormous
size of the IPv6 space - it also needs coordinated management.

> By introducing SDN/NFV in BNG, the

The problem exists without SDN and NFV. I agree that virtualization of
any kind makes the problem bigger, but it doesn't *create* the problem.

>    address pools can be allocated in a centralized way.

As we have already discussed, the *requirement* is coordination, not
centralization. In fact, again considering the enormous IPv6 space
within a carrier, a distributed solution seems unavoidable in the
long run.

>       APMS A management system which has a centralized databse manage
>       the overall address pools and allocate address pools to the device
>       in the devices.

Again, although a central database is the traditional approach, we must
be open to a scalable approach and that means distribution. Certainly
there must be an *initial* pool that originates by human action in
the NOC. But that does not require a central database that records
every detail. 

If an ISP has say a /29 and is delegating /48 to subscribers, it
has to manage up to 2**19 prefixes. That isn't an impossible size
for a central database, but it is large. (And this is a small ISP.)
The problem becomes much easier if you distribute it among, say, 100
agents spread around the network.

Apart from that comment I like your model with numerous agents
in the network. Actually it corresponds very closely to the ANIMA
model except that we will distribute the function of the APM server
among those agents. In my toy implementation I found that it was
necessary to define an initial "master" agent that provides the
initial pool - the difference from your APM server is that the master
is identical to all the other agents except for having the initial pool.
(However, I repeat that this is a toy, not a serious implementation.)

I look forward to discussions in Chicago.

Regards
   Brian Carpenter