Re: v6ops:IPv4 vs. IPv6 operational costs

[Tim Chown <tjc@ecs.soton.ac.uk>]

On Wed, Nov 23, 2005 at 01:21:25PM -0800, Fred Baker wrote:
> that would surprise me. My Mac is dual stacked now, and they tell me  
> that Windows Leghorn (Longhorn?) will be as well. Comes that way out  
> of the box.
> 
> The question will be whether IT departments turn it on in the  
> network, not whether the hosts will use it if hey do.

The question isn't only the 'cost', it's also about the benefit.  You 
also need to define 'cost' :)

Our experience of deploying dual-stack as far as open source and vendor 
provided solutions go on an enterprise of over 1,000 hosts is that the
additional 'cost' is low.   

We didn't buy new hardware specifically for IPv6.   We used to run a
parallel IPv6 routing infrastructure on BSD, but went 'properly' dual
stacked when we reprocured our old IPv4 equipment this summer.   No
extra cost per se.   The MRs on the procurement meant the new kit 
supported IPv6 features (unicast amd multicast).   The costs lie more in 
managing the additional protocol, and in ensuring support staff have 
expertise to do their job.

For most platforms, deployment Just Works.   The main gaps we have lie 
in certain vender applications (e.g. Exchange) and in commercial firewall 
product, but from the host and router platform view, the picture is good.
We have no issues running DNS, MXs, web, etc dual-stack.

We have occassional minor operational issues, but these cause very 
little extra supprot effort.

By deploying dual-stack early, we believe our staff gain the expertise in
IPv6 at an early stage.  As an educational site, we expose our CS graduates
to be to the technology.   That is a benefit in our context.

I would imagine the backbone NRENs that have deployed would say likewise,
provided they deployed IPv6 incrementally through new procurements 
rather than going out and buying new line cards specifically to support
IPv6 in hardware.   

You can find example deployment reports at www.6net.org.

In terms of benefits, well, restoring global addressability is important,
and in places where IPv4+NAT was used we can run IPv6 in parallel and
use point to point conferencing and other applications that would have 
been very hard to do between two NATed sites.   Dual stack with IPv4+NAT 
alongside IPv6 will be common, I believe.

Another benefit is that, as an educational site, we see new innovation and 
interest in networking, I believe in part because catering for NATs is no 
longer an issue.   A couple of student developed packages have attracted 
(some) commercial interest.

I think Fred sums it up well when he says that had a business case been
required for the web, it probably wouldn't have happened.

-- 
Tim/::1