Cisco First Out of the Box in the New 'IPv6'
Router Derby
By Jim Seymour
Special to TheStreet.com
12/27/00 11:35 AM ET
"IPv6." You probably haven't encountered that less-than-euphonious acronym yet, but
mark my words, you will. And soon.
It stands for "Internet Protocol version 6," and it's going to shake up the computer and
Internet hardware and software business, the Internet service provider and hosting
businesses, and the wireless business.
Which means we can make some money here. So bear with me for a minute as we
dig into the arcana of Internet Protocol.
The present version (v4) of Internet Protocol -- the IP we hear about all the time, as in
"IP address," or "IP telephony" -- has serious problems in a world gone Net mad. The
biggest is that there just aren't enough IP addresses available under the "IPv4"
scheme.
An IP address is the unique identifier assigned to a Net-access device during its Net
sessions. It takes the form of nnn.nnn.nnn.nnn, where each "n" is a single-digit
number. Under that plan, we have a maximum of 4 billion IP addresses, which sure
sounds like a lot.
Moreover, because most of us -- or rather, our computers -- don't have permanent IP
addresses, but rather use a different one drawn temporarily, every time we go online,
from a pool of IP addresses, those 4 billion theoretical IPv4 addresses go even further
than you'd think.
We use temporary IP addresses? Right. When you log on to a dial-up Internet service
provider, the ISPs' computers search a pool of IP addresses assigned to that ISP, and
give you an unused one for the duration of that session. Log off, then back on, and
you'll get a new, different one, ad infinitum. Since only a percentage of any ISPs'
dial-up customers are actually online at any given moment, these dynamically
assigned IP0-address pools have worked well.
If, on the other hand, you have an always-on Net-access device -- as with, for
example, cable-modem and DSL-modem service -- you need, and get, a full-time IP
address. That burns the available supply of IP addresses a lot faster, of course. Still,
with 4 billion IP addresses available, it seems as if there shouldn't be a problem.
Actually, there isn't -- yet. But a big problem lies right around the corner, and the Net
industry generally -- especially Net plumbers like Cisco -- is scurrying to position itself
for the advent of IPv6.
IPv6 is the product of what at times looked like an endless series of standards-setting
meetings. (There never was a viable IPv5; it got lost in the standards shuffle.) It
addresses many more issues than only the number of available addresses, but the
more-addresses issue is what's going to get all of the attention when IPv6 surfaces in
the public consciousness. Because the new IPv6 scheme provides for -- ready for
this? -- 340 trillion trillion trillion addresses. That sounds ridiculous, but it may not look
so funny in a decade or so.
The hidden issue is the coming flood of nontraditional Net devices -- all built around
the expectation of "always-on" full-time connectivity. And thus all of them are devices
that require assignment of a permanent, full-time IP address when they're
manufactured. From so-called Internet refrigerators to Internet home-alarm systems to
Internet-equipped automobiles to billions of pocketable wireless devices (I don't think
we'll be calling them "cell phones" much longer), we're going to be gobbling up an
amazing number of IP addresses.
Indeed, a friend of mine who has been working on this problem for a long time
estimates that within a few years, many of us will each effectively consume 100 or
more IP addresses. And, of course, as the Net continues to grow, new servers will be
added, too ... and every Net device, including those servers, will require its own IP
address.
So IPv6 is a bet on the future, yes -- but on a future not Out There Somewhere, but
just around the corner.
Virtually every computer and software maker has announced (but not delivered)
support for reading and using IPv6 addresses. Microsoft , for example, is hard at work
on IPv6 implementations, applications and utilities. While there is debate about how
well IPv6 will be able to integrate seamlessly with today's IPv4 addresses, I'm
optimistic that this will be a relatively smooth evolution. Embedding IPv4 addresses in
larger IPv6-style psuedo addresses is an inelegant but promising path.
Cisco is trying to lead the pack in the router business into the IPv6 Era -- appropriately
so, given Cisco's control of the router business.
Two weeks ago Cisco fellow Stephen Deering announced at yet another IPv6 industry
conference, this one in Osaka, Japan, that the company will finally begin shipping an
IPv6-translation router in 2001. He laid out a three-step release schedule, starting with
a mainly software implementation in February, which won't be feature complete; then a
more robust and complete, mainly hardware IPv6 router in midyear, focused almost
entirely on resolving IPv4-IPv6 issues; then a final answer, sweeping in lots of other
IPv6 improvements beyond just more address space, at an as-yet-unannounced date.
We can expect to see similar announcements from other Net hardware providers. And
look for all future releases of Windows to be IPv6 capable.
Internet service providers everywhere are gnashing their teeth over this evolutionary
move, caught in the usual conundrum between knowing it's necessary, and fearing its
cost. But the router people are going to sell a lot of new hardware, and a lot of
software upgrades, to make the IPv6 conversion happen.
This is also clearly a positive for the makers of non-PC devices, such as smart
personal digital assistants, smart phones and so on. The wiring has to be in place, so
to speak, before IPv6-capable devices catch on, so we'll see the usual tug of war
between early device adopters and perennially late service providers.
thestreet.com |
