On Tue, 06 Nov 2001 09:37:45 EST, Hans Kruse <kruse(_at_)ohio(_dot_)edu> said:
The discrepancy in opinions below seems to me to point towards the
deployment path for IPv6. Corporate users and those with very large
address space needs (wireless handhelds) will deploy IPv6 and in effect
"pay" for the engineering cost of building IPv6 into operating systems and
network elements. Once the costs come down, home users, small businesses,
and their ISPs will follow.
Actually, the engineering cost of building IPv6 into operating systems
is already essentially paid. The cost of building it into routers and the like
is paid. The vendors are all (basically) IPv6-ready.
What has NOT been paid is the deployment cost, which has several very
distinct components.
First is the cost of obtaining a release of your operating system
that supports IPv6 (which can vary anywhere from free to low-cost to
exhorbitant, depending on your vendor's business model, and whether
you're already running an IPv6-capable version for other reasons).
Second is the cost of installing that release - which can get complicated
if this forces an upgrade of third-party software such as a database
system. This has several components - additional licensing costs for
new releases of add-on software, downtime costs, testing costs, and
all the other little things like that. And add another 1% for each
time you ask yourself "Will this upgrade unexpectedly hose something
in a totally non-obvious way?".
(Note - none of the money so far has anything to do with IPv6 directly)
Third is the cost of actually configuring and enabling IPv6 - getting
an AAAA record assigned, the DNS set up, testing, and getting the software
to use an IPv6 connection. This is usually the cheap part once
you get past the first two.
Note that in the current state of the world, those first two are usually
the big part of the expense, and THOSE costs are *NOT* going to come
down anytime soon. That big database server on that big Sun E10K is
not going to be doing stuff over IPv6 until you upgrade to Solaris 8,
and drag Oracle along kicking and screaming. And let's be honest here -
at the low end, if you're a Microsoft user, unless you're technically
skilled enough to install the developer toolkit and roll your own,
you're not doing IPv6 until you upgrade to Windows XP - which means
you also get to buy a Norton upgrade, a this upgrade, a that upgrade.
The cost is *NOT* in deploying IPv6. The cost is getting to a
configuration that is *able* to run IPv6.
Valdis Kletnieks
Operating Systems Analyst
Virginia Tech