Slashdot is powered by your submissions, so send in your scoop


Forgot your password?
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. Also, Slashdot's Facebook page has a chat bot now. Message it for stories and more. ×
The Internet Communications Networking Technology

Van Jacobson Denies Averting Internet Meltdown In 1980s 57

New submitter strangebush sends this quote from Wired about Van Jacobson's work on the TCP/IP protocol in the '80s, which helped stabilize early computer networks enough for them to eventually grow into the internet: "'I was getting a bit per second between two network gateways that were literally in the same room,' Jacobson remembers. ... In 1985, Berkeley ran one of the IMPs, or interface message processors, that served as the main nodes on the ARPAnet, a network funded by the U.S. Department of Defense that connected various research institutions and government organizations across the country. The network was designed so that any node could send data at any time, but for some reason, Berkeley's IMP was only sending data every twelve seconds. As it turns out, the IMP was waiting for other nodes to complete their transmissions before sending its data. The ARPAnet was meant to be a mesh network, where all nodes can operate on their own, but it was behaving like a token ring network, where each node can only send when they receive a master token. 'Our IMP would just keep accumulating data and accumulating data for about twelve seconds and then it would dump it,' says Jacobson. 'It was like the old token ring networks when you couldn't say anything until you got the token. But the ARPAnet wasn't built to do that. There was no global protocol like that.'"
This discussion has been archived. No new comments can be posted.

Van Jacobson Denies Averting Internet Meltdown In 1980s

Comments Filter:
  • by AuMatar ( 183847 ) on Saturday May 26, 2012 @11:02AM (#40121291)

    I also did not avert an internet meltdown in the 1980s.

  • by file_reaper ( 1290016 ) on Saturday May 26, 2012 @11:55AM (#40121575)
    Here's Van Jacobson's Tech Talk at Google in 2006: [] I didn't know much about Van Jacobson's work on networking before that, I found it quite informative, thought I'd post it here.
  • Grammar in artlcle (Score:4, Insightful)

    by owlstead ( 636356 ) on Saturday May 26, 2012 @12:17PM (#40121767)

    It used to be that reporters first learned basic grammar before creating an article. English is not my first language, but that article has been written so badly that it is hurting my eyes. Even the quotes don't make sense (if they are actually quotes, who can tell?)

  • Sorry to all those who tried to be clever.

    Microsoft did not poo-poo the Internet. They just didn't "get it". That's par for their course.

    BSD did not create the TCP/IP software stack. They just had more runs at it and by the tie Reno and Tahoe came along got it almost right.

    Van Jacobson is a fairly smart guy. Pretty much if he says X then you can bet X is true.

    IMPs were around long before the Internet, the NSFnet, and only applied to ARPAnet.

    With all due respect to all of us who were working on networkin

  • So, we have the headline:

    Van Jacobson Denies Averting Internet Meltdown In 1980s

    And then we have the summary, which appears to recount a not particularly exciting anecdote about how a guy noticed things were going a bit slower than they should for an indeterminate amount of time and with indeterminate consequences. It doesn't even tell us what Van Jacobson did do when he wasn't averting a mythical meltdown.

    • by swalve ( 1980968 )
      That's why you read the article, Brainiac.
      • I should read the article because the headline doesn't relate to the summary? Not sure what your logic is there. A headline should be to a summary as a summary is to an article.
  • As one of the people who was active in TCP design back then (see my RFCs), this article sounds weird.

    First, the ARPAnet was not "the Internet". The ARPAnet was a closed backbone network, with flow control and guaranteed delivery of packets. When hosts talked directly to ARPAnet nodes (IMPs), the backbone provided reliable transport. When Ethernet to ARPAnet gateways were created, the possibility of packet loss in gateways appeared, and congestion packet loss became a problem.

    The TCP/IP implementation from Berkeley in BSD wasn't the first; it was about the fourth. We at Ford Aerospace used 3COM's UNET, which was a very early TCP/IP. I had to overhaul it, adding ICMP. UDP, congestion control (that's why I have those RFCs on network congestion), and checking for invalid packets. After that it could talk reliably over fast or slow links and to other valid implementations. We had a real "bit bucket"; all packets that didn't meet the spec were logged, and I used to check that every day and send out notes to other TCP implementers. Mark Crispin at Stanford was responsible for the PDP-10/DEC-20 implementation, and we talked a lot as we made two very different implementations play well together. I was impressed with Mark; unlike many developers today, he never blamed someone else when his end was at fault. I once sent a packet to Stanford which caused the implementation there to crash the mainframe, and I apologized to him. He wrote back that it was his fault if his mainframe crashed, not mine.

    The Berkeley people had originally assumed that TCP/IP would use Ethernet as a backbone and didn't worry too much about interoperability with other TCP implementations. Berkeley UNIX up to 4.3BSD could barely operate over a slow or congested link, and interoperated badly with other TCP implementations. The initial release of 4.3BSD would only talk to DEC-20 implementations for 4 hours out of every 8, because the sequence number arithmetic in BSD had been botched. (I had to fix that, which was a painful 3 days.)

    Van Jacobson was responsible for bringing the BSD TCP up to an acceptable level of behavior under heavy traffic. That was a few years later, around 1988.

    3COM discontinued UNET in the early 1980s, since UC Berkeley, funded by the Government, was giving away a comparable product. Ford Aerospace got out of networking because they only did DoD work, and networking was going commercial. I left Ford Aerospace, and networking, in 1986 because a friend of mine had started up a little company to do CAD software, and it was becoming successful.

    John Nagle

Machines have less problems. I'd like to be a machine. -- Andy Warhol