Subject: Re: (Somewhat OT) Re: INET6 in GENERIC
To: Thomas E. Spanjaard <tgen@netphreax.net>
From: J. Scott Kasten <jscottkasten@yahoo.com>
List: tech-net
Date: 02/22/2006 10:43:24
On Wed, 22 Feb 2006, Thomas E. Spanjaard wrote:
>
> Yup, the fact is the US still has a sizeable reserve of unused/unallocated 
> IPv4 addresses. However, besides the address space exhaustion issue present 
> in the rest of the world, IPv6 has other merits as well, which are a real 
> tangible benefit to all of us (some related to the huge address space, like 
> simple prefix routing because there's so many space). I think these merits 
> are underlit currently.

Routing, standardized IPSec, QOS, etc...  Many benefits besides just more 
addresses.

Before my comments below sound overly critical, let me say that I'm on 
board with v6 to be sure.  What I give you below is strait from the 
executive board room.  I'm always looking for arguments to support v6 
development in products.

> Yes, for instance, mobile devices connected to the Internet. The number of 
> mobiles in daily use is surely so large, IPv4 reserves can't cater for it.

It is the mobile phone generation that is driving this in Europe, China, 
and Japan.  Cell phones outnumber land lines in those countries and the 
full suite of 3G and 4G applications pretty much require each phone to 
have an IP address for delivery of video on demand, chat services, etc.

> Then, there's the booming internet media distribution market, which requires 
> media playback devices to have access. With that, there's numbers of those 
> 'business opportunities' that will thrive only if the Internet can sustain 
> the number of devices connected. It's easy to see the 32-bit IPv4 address

And that is precisely where the argument fails with most managers and 
executives.  They look at that and ask, "What does that have to do with my 
product?" (Assume printer, copier, fax, or other appliance.)  It becomes 
difficult to describe just what the value add is for simple products like 
that.  The execs feel like you are talking about the pie in the sky world. 
They are concerned about whether this is going to bring in additional 
revenue in the next product release when this is going to add $500,000 to 
the development cost now.  (Think two developers hacking code for months, 
a user inteface team spending months figuring out what easy 
configuration would look like then running teams of people through 
usability testing, an entire quality assurance team developing new 
regression tests, and spending months actually testing it.)

> space won't suffice. Now, there are technologies like NAT, but they're just a 
> hack on a standard that didn't cater for it. An ugly hack at that. Now for 
> all those cost-conscious 'decision-makers' out there, I'd like to ask if it's 
> worth throwing money at making applications work with NAT each and every time 
> you invent them, now, and in the future, on a regular basis? Or throw money 
> at a new standard once for a prolonged period? I think the choice is easy...

From the technical standpoint, I agree with you totally.  But an executive 
will "play it safe" by bolting something else on to what he already thinks 
he has rather than go invent something new.  Product sales is about 
variations on a theme - you don't mess with something that is already 
selling well.  NAT, how ever ugly it is, looks like a simple add on to an 
executive.  He's not the poor network admin that has to spend weeks 
figuring out how to rebuild the server room to make it work in real life. 
That's a hidden cost that is not easily measured.

For what it's worth, when I have input into the hiring process, I always 
try to get developers that have done at least some sysadmin work hired 
onto product development teams.  I feel that they develop better products 
when they actually know how it will be used.

Regards,

-Scott-