[rpd] Re: Factors affecting in-region utilization - way forward?

[Stephen Wilcox]

On 20 July 2014 20:01, Mark Tinka <mark.tinka at seacom.mu> wrote:

> On Saturday, July 19, 2014 04:24:00 PM Kofi ansa akufo
> wrote:
>
> > If an institution should go for such a service they
> > should not expect a /24 IPv4 or higher to be assigned to
> > them.
>
> I don't believe in promoting old stereotypic situations,
> even though they may be typical.
>
> Things are changing a lot faster than we've been accustomed
> to, and our industry has suffered a great deal because we
> like to keep things the way they've always been.
>

We are also scaling a system that was designed 30 years ago not to do what
it does today. Breaking the stereotypes could prove to be problematic -
core routers handling more than a few Mbps will switch in hardware and
rapid growth in the number of routes in the global table could quickly
overwhelm certain in service devices. Its okay to understand your 500k FIB
limited box has a lifetime of 3yrs when you buy it but if you see
irresponsible deaggregation or sudden announcement of small blocks you
might cause people's equipment to become EOL faster than they can cope.


>
> > Lesson: most internet users are only concerned about
> > being connected and how fast their internet connection
> > is. Only a few care to know whether they are NATed or
> > have a direct transparent IP connection. This is
> > universal.
>
> What is interesting is that NAT has the potential to slow
> down your performance, as you scale up bandwidth.
>

Is there data on this (I am not doubting you just that I haven't heard this
being stated before)

Steve


>
> Mark.
>
> _______________________________________________
> rpd mailing list
> rpd at afrinic.net
> https://lists.afrinic.net/mailman/listinfo.cgi/rpd
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.afrinic.net/pipermail/rpd/attachments/20140720/c10a51bf/attachment.html>