Chapter 9 details a real event which I actually went to (“Dan” is me, basically): the Internet Engineering Task Force meeting in November, 1991. Even though the general public was taken by surprise by Netscape Day in August 1995, the Internet had been getting built and tested for decades. Some companies were onboard, mostly smaller ones, while the larger ones like Oracle (where I was) fully intended to support it as soon as they could. Many of the attendees who did actual work were from academic or research institutions.
I tried to show how the “smart money” in Silicon Valley was not at all persuaded that this new Internet thing was ever going to take over. The world of networking then was dominated by proprietary networking protocols: Novell’s, IBM’s, DEC’s, 3Com’s, and numerous others. All of those companies claimed that their stuff was “standard.” One of Bob Metcalfe’s favorite quotes was:
The nice thing about standards is, there are so many of them.
Transmission Control Protocol / Internet Protocol (TCP/IP) was being developed for the Internet, and it was not owned by any one institution or corporation. Nor was it limited to any one vendor’s equipment.
I joined Oracle in March, 1991, and I distinctly remember our VP, Smokey Wallace (RIP) say, paraphrasing here:
All those protocols, SPX, DECNet, SNA: those are not going away. Everything is not going to be TCP.
Very soon after that, everything was TCP.
What Was Bob Doing?
As Dan says:
I’m doing this database to help people manage all the text files you need with Unix and the Internet. Learning all about /etc/passwd (he pronounced it “etsy-password”), DNS files, and all that.
That was it. Setting up Unix and TCP required a lot of error-prone text editing, so naturally Oracle wanted to create a bunch of forms with a database, to enforce some rules and consistency.
Oracle was “agnostic” on networking: “whatever the customer wants to use to get at the database, we’ll support it.” SQL*Net was an extra-cost option once you’d bought the database, and almost all the customers bought it. That meant that our division had immense revenue coming in! That’s generally a good thing, although it did mean a lot of pressure from the rest of the company.
Platforms
This part is a little geeky. Skip over it if you’re not a programmer.
Oracle was shipped on about 90 platforms. It’s hard to believe there were so many, since now there are only a few. There were a lot of hardware companies (Sequent, Pyramid, HP, IBM) who each had their own version of Unix, all subtly different. Oracle had a set of coding standards for the “C” language, basically the lowest-common-denominator of all the platforms’ C compilers.
That meant, for example, that externally visible names were limited to 6 characters (later expanded to 8) because that was the rule on one compiler. And that, in turn, meant that name prefixes were centrally controlled at Oracle. No one could name an external variable ‘nsFoo’ because our division owned ‘ns’. There was an entire manual telling you things you could not do in portable code, e.g. assume an integer was 32 bits, or pointer arithmetic.
Each platform had a “platform group” whose principal task was to take the portable (supposedly) code shipped by the product groups, like the database kernel, and adapt it to their platform. The HP group was the most influential of the platform groups, since we sold the most Oracle software to HP customers.
Traveling around
As someone who wrote “portable” code, it was a normal thing for some platform group to complain, either in writing or by phone, that my code didn’t work on their platform. Normally I’d have to go there and sit down at their keyboard (this was before COVID, of course).
I’m an emacs
bigot, but emacs
wasn’t present on every platform, so at least I had to learn how to get out of vi
. (I guess people had started calling it “vim
” by then but I just called it “vi
”) That’s still all I know how to do.
SNMP
Simple Network Management Protocol (SNMP) is a standard for managing networks, and meetings about it were being held at IETF meetings then. I’d started doing that at 3Com, and it was just taking shape at Oracle. I loved it.
I ended up being in charge of Oracle SNMP, and I led a Working Group to create the RDBMS MIB (Management Information Base): RFC 1697. Since Oracle was the leading database at the time, it sorta made sense for Oracle to lead the group, but it was absolutely essential that the MIB was fair to everyone and didn’t give Oracle any special advantages. We had all the major databases vendors there and got the RFC written in about nine months.
Pornography on the Net
The encounter with “Stan” the pornography entrepreneur did not actually happen to me. A friend, Chris, met a guy much like Stan, who startled him by announcing the bright future of pornography on the Internet! The net was so slow then that it was hard to imagine it. Back then, video stores had a special section for adult movies on VHS.
I showed Chris that section and he said I captured it pretty well.
Dan and Len
We’re setting up a future friendship between Dan and Len, Janet’s father. This will be important later.
GO and Momenta
A little more abuse for Momenta, just to finish up with last week’s.
Santa Fe
Have I mentioned it was freezing there?
There are indeed hundreds of art galleries. Like everywhere else, galleries are a terrible place to buy art.
re: "I tried to show how the 'smart money' in Silicon Valley was not at all persuaded that this new Internet thing was ever going to take over."
During my short tenure at the Xerox product group, I noticed a bit of a culture gap between XNS and Internet folks: the Internet folks expected that if, say, listname@xerox.com existed, then listname-request@xerox.com would go to the list owner. The XNS convention was instead (if I remember correctly) to use owner-listname@xerox.com. So I submitted a feature request to automatically map listname-request to owner-listname, to avoid bouncing email unnecessarily. It got rejected. Part of the reason given for the rejection was something like "that's just ARPAnet, who cares about them. Our stuff is so much better."