Ready for the Next Wall?

In a couple of months, we IT professionals will be breathing a collective sigh of relief as the most publicized deadline of our time -- the dreaded Year 2000 rollover -- will have come and gone. If our billions of dollars of effort pay off, there'll be nothing to do come Jan. 2, 2000, except toast our professional triumph and maybe mop up a few forgotten Y2K-bugged COBOL programs.

But we better enjoy our post-Y2K party while it lasts; the light we're seeing at the end of the millennium tunnel might be from the oncoming train of the Internet's collapse.

Okay, maybe I'm overstating my case a little, but the fact is we're all headed for a major network reckoning: The Internet is about to run out of IP addresses.

It's true. IP's addressing scheme is reaching the end of its lifecycle. Designed almost three decades ago to accommodate a few thousand interconnected computers, today's Internet is using yesterday's IP to handle the interconnection of its many millions of subscribers, and it's just not up to the job. IP's familiar two-level class addressing system -- defining three groups of addresses for small (Class C), medium (Class B) and large (Class A) organizations -- results in extremely wasteful address allocation and painfully large routing tables.

If it weren't for some just-in-time technical slight of hand in the mid-90's, IP and the Internet would be dead by now. The introduction of the classless interdomain routing protocol (CIDR), coupled with the creative assignment of aggregated groups of available Class B addresses to ISPs, was able to buy us some time.

But not much time. Depending on projections for future Internet growth, CIDR may run off the rails anytime between 2002 and 2008.

Fortunately IP's replacement is waiting in the wings. A new version of IP, IPv6, also called IPng for IP Next Generation, has evolved during the past several years into a draft protocol standard, and has been implemented -- in both end-node and routing configurations -- by several vendors.

Its implementation, unfortunately, won't be transparent. We will all have some work to do. While it is designed to be implemented incrementally, running in parallel with today's IP, we will eventually have to retool every IP-speaking machine on our networks to take full advantage of IPv6. And as we depend more on IP and the Internet, our future upgrade task will become more daunting.

So what should you do? Start by learning as much as you can about IPv6. You will see that, beyond its resolution of the Internet addressing problem, IPv6 offers multicasting improvements, enhanced address administration and better throughput. A good IPv6 technical summary can be found at the National Communications System's Web site (www.ncs.gov/n6/content/tibs/html/tib97_1/sec2_0.htm), and a host of informative IPv6 links are listed at the protocol's unofficial home page (http://playground.sun.com/pub/ipng/html/ipng-main.html).

Armed with some of this information, you can begin taking a general inventory of the affected devices in your network. Think about it. Eventually you'll have to implement IPv6 in all of your Internet-equipped PCs and servers, your IP-equipped routers, your OSI-level-three LAN switches and your DHCP, domain name service and other TCP/IP protocol support boxes. And don't forget your personal information managers, such as Palm Pilot and Windows CE appliances -- many of these use IP addresses, too.

Remember 2002 is not that far off. Not when you consider all the firmware upgrades, software installations and router configurations that will need to migrate to IPv6. Go ahead, toast your Y2K success and drain that glass of champagne. When you're finished, toss your empty glass into a nearby fireplace and start visiting those IPv6 Web sites. Forewarned, they say, is forearmed. --Al Cini is a senior consultant with Computer Methods Corp. (Marlton, N.J.) specializing in systems and network integration. Contact him at al.cini@computermethods.com.