Neat Net Stories from Vint Cerf and Dan Lynch at Interop

A Better Migration Plan to IPv6?

Interstellar networks. Padding Statements of Work in US DoD contracts to encourage the development of the Internet. Turning off the old protocols on the Internet, for a day, to encourage the adoption of the "new" TCP/IP protocols.  The plan to use OSI layer 3 protocols (well, kinda) instead of IPv6. All these topics, and more, were part of an interesting pair of discussions at Interop yesterday with Vint Cerf and Dan Lynch, two pioneers of the Internet and the networking industry. Today I'll pass along a few of the tidbits that I personally found interesting.

As part of the keynote session Tuesday AM at the show, Vint Cerf, currently Chief Internet Evangelist for Google, and Dan Lynch, founder of the Interop show (in its 25th anniversary year this year), sat on the stage and discussed the history of the Internet and some ideas about where we are headed. Both men played important roles in the formation of the Internet, and were there when it all happened. After the session, they took questions from the media as well. And while this is a departure from what we normally discuss here, it was my favorite part of the show Tuesday, and I found it interesting, so the rest of today's post is a ramble of what they said, or at least my remembered version of it.

First, as a quick history review, the Internet grew out of some US Department of Defense (DoD) contracts. The predecessor protocol before TCP/IP was called NCP. In the 1980s, the Internet didn't exist as "the Internet", but consisted of various networks, including ARPAnet (Advanced Research Projects Agency) and MilNet (the US military equivalent), with gateways (routers) between them.

Now for some random unrelated remembrances:

Vint: IPv6 adoption is the key to the future growth of the Internet.

Dan: They needed load to help test ARPAnet. One of the early big loads? A mailing list (in the way early days of email) about "Mary Hartman, Mary Hartman", a way cheesy late night soap opera.  Not exactly DoD related work; but it got some response like (paraphrased): "We don't approve of the content, but that's how you get load".

Vint: He used to add wording to most Statements of Work for DoD projects - something widely phrased - just so the folks doing the work could try anything to exercise the network.

Both: there was agreement that to replace IPv4, they would use parts of OSI's layer 3. (I was unclear as to specifically what they would use from OSI layer 3, by the way.) Regardless, the big idea was that instead of recreating the wheel, they'd use the work already done for OSI. Then there was the naysayers, people getting on Vint's case in particular, since of course OSI and TCP/IP folks didn't get along in those days. So Vint show up at a meeting with his usual three-piece suit, strips down to his undies, to reveal his "IP on everything" t-shirt. And that's part of the story of how we ended up with IPv6.

Vint: ARPAnet and MilNet - think early Internet, two parts, connected with routers. One part for research, one part production but protected, for the US military. Someone had the great idea to disconnect the two from each other, to help secure the MilNet side. So they did. Who suffered? The change failed a payroll app on the MilNet side that relied on data that was on the ARPAnet side.

Vint: while at MCI in the 1980s, he helped create an email service called MCI Mail. Well, it was a nice idea, but the "Internet" at the time only allowed folks attached to research projects to connect. It certainly wasn't commercial. So... how to break that logjam? How to open it up to commerce? Well, in 1987, he asked, hey, can I do this MCI Mail service through the Internet, to test? The powers that be said yes. Then several other companies jumped in and said that wasn't fair, and they got connected, and then the snowball was headed downhill towards a commercial Internet.

But my favorite, and longer, story is one that Dan and Vint told together about the migration to TCP/IP, and IPv4, from the earlier NCP protocols.

Dan's group (I believe it was at USC) started pulling stats to see how many of the hosts on the net had added support for TCP/IP, in anticipation of the day at which they were literally going to disable NCP on ARPAnet and force a migration. ARPAnet  had an easy-to-use NCP on/off switch. After a while, Dan's stats showed no growth in the number of hosts implementing TCP/IP, so the rest needed a nudge.

So, for a day, Vint had NCP disabled. People screamed about it. But guess what? Dan's stats of hosts migrated to TCP/IP went up again... for a few months, and then stopped. This time, Vint turned off NCP for 2 days. A few months later, they made the cutover from the old protocol suite, to the new TCP/IP, with only 2 hosts not being ready for the big day. (There were about 400 hosts on the net at the time.)

So, maybe we need a migration plan like that for IPv4 to IPv6. Turn off IPv4 for... an hour? So I asked Vint about it at the press conference, and the serious answer was that he thought it would be a bad idea, just due to the damage. He's probably right. But it would sure speed migration, don't you think?

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2011 IDG Communications, Inc.

SD-WAN buyers guide: Key questions to ask vendors (and yourself)