Re: [v6ops] I-D Action: draft-ietf-v6ops-ipv6-ehs-packet-drops-00 - Admin Policy

[Fernando Gont <[e-mail]>]

Hello, Eduard,

On 3/8/20 10:03, Vasilenko Eduard wrote:
> Hi all,
> I do not believe that EH would ever be "for free" - it would consume PFE cycles (performance, energy).

In principle, I agree.

The Internet architecture assumes that intermediate systems only need to 
look into the destination address of a packet to make a forwarding 
decision. In such (theoretical) architecture, EHs would be "for free" -- 
or, at the very lest, free for the network (and only possibly expensive 
for the endpoints).

However, the reality is that intermediate systems quite often do need to 
look at layer-4 values to make a forwarding decision. Therein lies the 
problem -- EHs are not "free" anymore.




> IMHO: "EH success" is the situation when small set of EHs would have low drop rate.
> IPv6 Architecture should be friendly for this to happen.

The variable-length daisy-chain-like structure of IPv6 packets is, by 
definition, not friendly to intermediate systems that need to find 
layer-4 information.

That's why -- besides possible filtering policies -- EHs are dropped.

Thanks,
-- 
Fernando Gont
e-mail: [e-mail] || [e-mail]
PGP Fingerprint: 7809 84F5 322E 45C7 F1C9 3945 96EE A9EF D076 FFF1