Pekka Savola wrote:
[...] you seem to have a fundamental misunderstanding of the context of the word "send" in the rate-limiter specification [...]
It is amusing for you to say that. As a co-author of the ICMPv6 RFC 2463, and RFC 1885, work which started in 1994, I have a good insight on the meaning of the text, and the meaning of the standard. However I am not interested in giving lectures, and am not interested in proving how wrong you are in interpreting the ICMP specs. I am rather interested in a quick positive outcome and a win-win situation.
Are you ready to be constructive?
If your concern, your vision and your target is "controlling internal ICMP message generation" and the mechanism to achieve that is the token bucket, I do not understand why does it matter to you the number of token buckets, whether it is one, or many?
I hope you understand that with one token bucket per node, you cannot control ICMP packet generation in a modern router, so basically, routers
are left out.
Just to make sure we're on the same page:
in a modern router architecture there is distributed packet forwarding, i.e. multiple line cards, each line card with its own implementation of link, IP, and ICMP processing, as well as traffic management, and support for multiple interfaces, that have bandwidth which ranges from fractions of 1.5Mbit/sec at one end of the spectrum, and 1Gbit/sec, or 10Gbit/sec at the other end of the spectrum.
So you understand now that because each line card may have at least one IP/ICMP engine, the token-bucket per node cannot apply?
Therefore, the current recommendation in the ICMP specs should be upgraded to support rate control per line card. As a line card has several interfaces, in general of same type, it is sufficient to specify the rate control for one of the interfaces, which brings us to the interface based rate control configuration, and the one sentence text addition to the ICMP specs that I suggested.
So, in order to support distributed control of ICMP message generation, the addition to the spec which I suggested is necessary, since one token bucket per node cannot apply.
How is this with you?
smime.p7s
Description: S/MIME Cryptographic Signature-------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
