Comments

16 responses to “Can any old-timers from #SunMicrosystems please pass comment upon “Commercialization of TCP/IP” / Martin Leufray”

1. 2012/05/02

   Brad

   This hardly worth the time to refute. This paragraph on particular is hysterical:

   “In 1989, I joined DHL. At the time, the world of telecom was in a huge war between ISO protocols (the international standard) and TCP/IP (at the time, only a US standard). It was the entire world against the US and it was based on the hatred left over after the breakup of AT&T in 1934. Overseas, they hold grudges literally forever. AT&T used to own all the phones in the whole world. There was no other telco. Although the international monopoly was broken up because the whole world was screaming at the top of their lungs, the national monopolies survived. The AT&T monopoly in the US lasted until 1984, 50 years later, when they signed yet another Consent Decree as part of a settlement with the US government and IBM, who had been fighting AT&T for decades.”

   Not to cite Wikipedia as the be-all-end-all authority, but look at the entries for IT&T vs. the entry for American Telephone & Telegraph (not AT&T, which describes the modern company that emerged from Southwestern Bell).

   American Telephone & Telegraph (much less AT&T) never had an international monopoly. They never had much of an international presence at all.

   I’m assuming the rest of his rant is equally bogus.

   For the record, I do have 1988 USENET postings from folks at MITRE and Convex from with company ARPA/Internet addresses.

   Reply
   1. 2012/05/02

      alecm

      Back in 1992 when I joined Sun, the internal network (SWAN / Sun Wide Area Network) was approx 20,000 IP-connected interfaces, some multihomed; by the time I joined Sun’s Network Security Group in ’94 it had grown to about 33,000 IP commections. I know this because I was allowed to hack them, and because of AutoHack I had possibly the most comprehensive picture of the entire network in the entire company.

      It’s not quite as early as this guy is apparently stating, but it was sizeable and back then we were referring to it as the second- or third-largest IP network in the world, after the Internet and (maybe) MILNET.

      And as for tech companies? Sun, Pyramid, Convex, Sequent, DEC (though they were mostly DECnet to be fair).

      I don’t think this chap makes his case on that front, either.

      Reply
2. 2012/05/02

   Stephen Smoogen

   Wow.. I hate to say this but I am not sure that fellow can get in and out of the car with brass balls that big.

   Of course the fact that the mag tapes are probably unreadable means this is unverifiable due to lost data :).

   Reply
3. 2012/05/02

   alecm

   from FB:

   Bob Douglas: Compuserve was offering online access at least by 1987.
   03/15/1985 SYMBOLICS.COM
   03/19/1986 SUN.COM

   Reply
4. 2012/05/02

   alecm

   from FB:

   Carole Fennelly: [it] belongs in the same category as the guy who claimed to have invented email:http://www.washingtonpost.com/blogs/omblog/post/reader-meter-who-really-invented-e-mail/2012/02/24/gIQAHZugYR_blog.html

   Reply
5. 2012/05/02

   Clive

   For the most part, it’s possible to draw important conclusions about that article’s objectivity and veracity from its tone. I especially liked the bit about “I probably still have that email on mag tape in storage”. Given that little old me knows how to carry the e-mail archive forward from each computer system to the next so it remains online, I would have hoped world’s best network architect would do the same.

   When Sun’s finished with this clown, Tim Berners-Lee, Phil Karn, Cisco, FTP Software etc. would probably also like a word with him.

   Reply
6. 2012/05/02

   alecm

   Also: seems his stellar career at EarthLink lasted 21 months in total, according to linkedin.

   Reply
   1. 2013/03/11

      Catherine

      Minor detail, but while he may have technically worked for the company longer, Martin was in charge of technology for less than year, about 10 months from Spring-summer 1995 until February 1996, when he was fired from that job.

      Reply
7. 2012/05/03

   Kate Stout

   Define “commercialization”. Both Sun and FTP Software were shipping commercial software that included TCP/IP stacks long before the founding of Earthlink in 1994. I know I was working on PC-NFS, which included a TCP/IP stack for Sun back in 1989. MS-Windows shipped NT 3.1 with a TCP/IP stack in 1993.

   And as someone mentioned, his AT&T info is all wrong. AT&T (a US company) was split into the “Baby Bells” in 1984. This had no impact on international phone systems. For example, the UK had BT, France had France Telcom, etc.

   Reply
8. 2012/05/03

   Henk

   I joined Sun this month in 1990, and it had world-wide connectivity. I worked on the helpdesk and had frequent conversations with guys in the US, across SWAN. We had email, nfs (udp across the pond!) and ftp. We even had an internet proxy (was that port 3666?)

   Before that I had been in touch with Geoff Arnold on some PC-NFS stability issues (*and* a bug in the SunOS kernel that could crash on malformed NFS requests).

   So, Sun was well connected in the late eighties.

   Reply
9. 2012/05/03

   diamond

   I worked for AT&T in the 80’s. At that time, I was involved in transitioning AT&T’s UUCP-based dial-up and X.25 internal networks to TCP/IP over Frame Relay. We were well connected with Universities, other companies, etc. long before this guy came around.

   Reply
10. 2012/05/03

    Geoff Arnold

    For a good picture of the vibrant commercial TCP/IP marketplace in the late 80’s/early 90s, check out the (mostly accurate) Wikipedia entry on WinSock. Not only does this show the number of vendors; it also conveys the urgency of the need for standardization, which was driven by the rate of new application development. (We were not all building the same four apps over and over again!) In retrospect, the timing was perfect, we came out with WinSock 1.1 (the first really usable version with widespread implementation) in January 1993, and a few months later the first Windows version of the NCSA Mosaic web browser emerged.

    But couldn’t all of this stuff have been restricted to a small .edu, .gov or .org clique? Nope. The most obvious example is RFC 1041, which dates from 1988. We tend to forget that one of the most important categories of TCP/IP-based applications back in the day was TN3270, a Telnet client that could emulate the IBM 3270 block-mode terminal. Long before 1992, a large number of companies had realized that a DOS or Windows PC with a 3270 emulator was a much more flexible and cost-effective approach to mainframe application access than a hardwired 3270 cluster. Some of the early PC TCP/IP vendors – NetManage and Distinct come to mind – specialized in this area, and they were selling to Fortune 500 corporations, not just research labs.

    Perhaps Martin Leufray III’s memory is failing him. All of this was so long ago. That’s the most charitable explanation I can come up with….

    Reply
11. 2012/05/03

    Carole

    I also worked for AT&T in the ’80’s (Bell Labs Comp Center, Holmdel) and not only were we connected all over the world (anyone remember the fun of updating your uucp “systems” file”?), we also were dealing with security issues. Hell, we even had a Corporate Security Task Force with a Sun working sub-group (which I was on along with Steven Bellovin) that talked about issues in NIS and the r-services.

    Reply
12. 2012/05/03

    Carole

    This may be picky, but he also says that 5ESS switches ran UNIX. They actually ran DMERT, which was based on Unix, but really had a different user interface. Kinda like saying DoS is based on Unix. Not a huge deal, but no one at the time would have called it Unix (I ran packet switches for the #1PSS project, aka “TNET” which was an offshoot of 5ESS. )

    Reply
    1. 2012/05/03

       alecm

       I wish I had a ‘like’ button. 🙂

       Reply
13. 2012/05/03

    alecm

    Just in case it goes away, I am going to quote Aleksey Tsalolikhin’s blog post in full, so we know what the above is in reference to 🙂

    Commercialization of TCP/IP – an account by Martin Leufray III
    Posted on May 2, 2012 by Aleksey Tsalolikhin

    Martin Leufray III kindly allowed me to re-post the account to an EarthLink alumni group of his involvement in the commercialization of TCP/IP:

    I joined EarthLink Network in April, 1995, as Manager of IT. I was soon promoted to Director of IT, at the time, the highest technical title in EarthLink. There was no CTO, yet. A CTO was not assigned until the pre-IPO madness.

    Although Sky very seriously doubted the practicality of my intentions at the time, I only agreed to join EarthLink to dominate the telecommunication industry worldwide. This is the much-abbreviated story of why that was a practical goal:

    By 1995, I had been in the industry for 20 years. My first 10 professional years were as a mainframe programmer. In 1985, I joined AT&T Information SYstems to help them sell UNIX boxes to major accounts, all of which only bought from IBM. AT&T’s policy at the time was that all starting employees were required to spend at least 25% of their time in class. I took a class on the Public Switched Telephone Network (PSTN). Bo-ring…

    That was until they started discussing the 5ESS (Class 5 Electronic Switching System), the biggest, baddest switch in the world. They said the switch ran UNIX. (UNIX pre-dates Linux by 22 years. Linux wasn’t invented until 1991, six years later.) *That* got me excited! I suddenly understood that all communications networks on Earth, the modern ones at least, were all based on regular, everyday computers running UNIX, and I understood computers like nobody’s business and I understood UNIX and the C programming language. In one moment I turned from a master of mainframes to a master of networking.

    In 1989, I joined DHL. At the time, the world of telecom was in a huge war between ISO protocols (the international standard) and TCP/IP (at the time, only a US standard). It was the entire world against the US and it was based on the hatred left over after the breakup of AT&T in 1934. Overseas, they hold grudges literally forever. AT&T used to own all the phones in the whole world. There was no other telco. Although the international monopoly was broken up because the whole world was screaming at the top of their lungs, the national monopolies survived. The AT&T monopoly in the US lasted until 1984, 50 years later, when they signed yet another Consent Decree as part of a settlement with the US government and IBM, who had been fighting AT&T for decades.

    I was hired by DHL to decide between ISO and TCP/IP for what was going to become the world’s largest network – DHLNET.

    I studied the situation for six months with no solution. Then I attended the 4th Interop, which at the time was basically focused on interoperation of TCP/IP networks and ISO networks. In a lecture by Dr. Marshall Rose, who was discussing his new book, “The Open Book : A Practical Perspective on OSI”, I finally grokked what he was saying — ISO was being implemented by various competing parties from a theoretical spec, so could never converge to an interoperational implementation. TCP/IP was based solely on interoperating implementations, with the spec being developed *after* and communicated by RFCs as documentation.

    I decided on TCP/IP for DHLNET and wrote a paper announcing it to all DHL technical offices. People flew in from around the world to try to convince me I could not make such a decision. They soon learned that I not only had the power to enforce my decision, but they would either go along with my program or get run over.

    I was invited to speak at the 5th Interop in 1990 as part of a panel with AT&T and HP, the 2nd & 3rd largest networks on Earth, to discuss why each of us were running which protocols on our networks. AT&T and HP both described their ISO-based networks and why they had decided against TCP/IP. I described our rapidly-growing and quite successful TCP/IP network and why ISO could never work in the end. After my presentation, both AT&T and HP said all they needed was for someone to take the plunge (me) and they would be free to go with TCP/IP. Both networks were changed to TCP/IP soon thereafter.

    So, you might think I am bragging, but I was responsible for the commercialization of TCP/IP and the technical infrastructure of the commercial Internet. In 1990, only the government, universities and commercial labs were operational members of the Internet. That all changed immediately after my presentation.

    As Chief Internet Architect, I rolled out DHL’s internal “Internet” to nearly every country on Earth. DHL owned over 80% of all commercial shipping on the planet at that time. FedEx only operated in the US and was only about 5% of the market.

    After I was done there, I started working for a startup in “Silicon Valley,” and I had the sweetest job on Earth — I was in charge of all IT and I had no limit on my budget. I could spend as much as I wanted. I was in Fat City and I wasn’t going anywhere.

    That is until Brian Wenger sent me a simple email: “Do you know the world’s best network architect?” To which I replied, “You’re talking to him. What do you need?” I probably still have that email on mag tape in storage. I have kept every email I ever sent or received since 1985.

    Soon, Sky and I were exchanging emails at a fast and furious pace, negotiating my terms of employment. For youngbloods who don’t understand why old school people use email so much, we want a permanent record so we have evidence in court in case a dispute arises. Email has been around almost 40 years. Email records have saved many an ass & many a career and probably billions of dollars in lawsuits.

    When I came to EarthLink in April, 1995, Sky was openly announcing in our weekly staff meeting that he had a dream of reaching 2,000 subscribers.

    Well, we kinda went a little ways past that

    Martin Leufray III
    29 April 2012

    [EarthLink reached over 5 million users. –Aleksey]

    I think he stands corrected; what Google will do with subsequent searches on Martin’s name is anyone’s guess.

    Reply