ietf
[Top] [All Lists]

Re: Deployment vs the IPv6 community's ambivalence towards large providers

2000-08-23 08:10:02
At 05:03 23/08/00, Rahul Banerjee wrote:

Well, not really just the enlarged address space! There do exist certain
other definite benefits of IPv6 like possible use of Flow Specifications in
the Flow Label, decreased processing delay at routers, greater flexibility
etc.  This is not to say that the IPv6 is answer to all problems, but just
to signify the worth of the effort of those working on it.

        I've tried really hard to avoid this debate, because it is the
wrong forum and it hasn't been a particularly useful debate.  The signal
to noise ratio has been remarkably poor.

        The erroneous information quoted above is part of the frustration
of the folks, like me, who aren't zealots either way and are just
trying to grow The Internet.  Equally erroneous information emanates
regularly from certain anti-IPv6 zealots, so there are no saints here,
only sinners.  However, I've written code for no less than 3 IPv6
implementations, thus far, so I like to think that I'm plausibly 
well informed.

Point by point:
        - Until a spec exists that explains how to actually use the Flow
          Label (and implement that use in Verilog or software), it 
          is NOT a "definite benefit of IPv6".  It is a *potential* 
          *future* benefit of IPv6, but is not a "definite benefit" 
          at present.
        - Other than cisco, most router vendors implement their IP 
          forwarding in hardware, so the forwarding latency for IPv6
          is really identical to that for IPv4.  In the near term,
          many vendors have ASICs to forward IPv4 but rely on conventional
          CPUs and software to forward IPv6 -- in this situation IPv6
          forwarding is HIGHER latency than IPv4.  In the narrow case
          where a vendor uses software-based forwarding for both IPv4
          and IPv6, it is highly implementation-dependent which, if
          either, is lower latency.
        - Greater flexibility.  Too vague a claim to evaluate either way.

        I think it would be helpful if the advocates toned down their
marketing AND if the anti-IPv6 zealots toned down their anti-marketing.
Neither is helpful, IMHO.  It would be useful if people on either side
could have a calm discussion about reality over a meal or beverages.
The last time I was present at such, the results were decidedly mixed.

        Sigh.

Ran
rja(_at_)inet(_dot_)org





<Prev in Thread] Current Thread [Next in Thread>