| Date: | March 1, 2016 / year-entry #47 |
| Tags: | history |
| Orig Link: | https://blogs.msdn.microsoft.com/oldnewthing/20160301-00/?p=93101 |
| Comments: | 30 |
| Summary: | It predated the Internet, you know. |
Commenter JT seemed quite astounded that Steve Ballmer sent email. There was no explanation as to the source of the astonishment, but commenter Boris thinks it was surprise that email even existed back then. Email has been around for a really long time. RFC 822, which spells out the format of email messages on the ARPA Internet (as it aas then known), was submitted on August 13, 1982. And it was itself a revision of RFC 733 from November 1977. Like most companies that used email back in the early days, Microsoft ran a closed email system that was not connected to the outside world. To read your email, you went to your Wyse 50 serial console terminal and connected to a machine running Xenix. From there, you logged in and ran the the classic Unix command-line mail program. (These terminals would later be repurposed as debugging terminals.) Having to switch to another keyboard to check your email was rather cumbersome, so two developers wrote a mail client that ran on your PC, connected to your Xenix server, signed in as you, downloaded your incoming email and uploaded your outgoing email. It also deleted the mail from the server after downloading, so you didn't get angry messages from the email administrator about going over quota. The incoming mail could be organized into mailboxes, so you could have a "Meetings" mailbox and a "Design" mailbox, for example. (No nested mailboxes. We're not that fancy.) The program was called WZMAIL (pronounced "whiz-mail"), named after its authors Mumble Wmumble and Mark Zbikowski. (Sorry, other guy!) J Allard described wzmail as "a time machine that was hardwired to 1982 BBS messaging systems."
Anyway, where was I?
Oh, right.
Like most corporate email systems at the time,
it was unconnected to the rest of the world.
At some point, somebody established an external email gateway
that used the UUCP protocol and routed through
the University of Washington's mail server named Bonus chatter: The term Information Superhighway as originally coined did not refer to the World Wide Web, although that's what it eventually turned into. At the start of the project that eventually became Microsoft Exchange, they chose X.400 as their model, seeing as X.400 positioned itself as the industry standard for email interchange. (You can see for yourself how well that worked out.) The product included a gateway that connected an Exchange-based network to an SMTP-based network. I remember filing a bug against the Exchange team pointing out that their gateway was generating headers that did not fully comply with RFC 822. The bug was resolved back to me with the remark, "We are not yet committed to supporting RFC 822 in our Internet gateway." Well, you can see who won that battle of the standards. |
Comments (30)
Comments are closed. |
Internet email was still transitioning from walk to run when I was in college in the very-early 1990’s. We had SUNY-wide DECnet email between SUNY schools, and globally BITNET was our option until AT&T finally got a fiber line to the out-of-the-way in-the-hills town with two universities. Someone (WUSTL – Washington University St. Louis – I think) ran a gateway called BITFTP that let you get files on FTP servers via chunked-up UUencoded emails. (http://docstore.mik.ua/orelly/unix/ksh/appc_03.htm)
Kids these days have it too easy.
P.S. – Who remembers ARCHIE servers?
Another example of where sometimes it’s not necessity that’s the mother of inventions, but laziness :-) And I’m ok with that.
Zbikowski managed to get his initials in quite a few places, didn’t he? I remember posing a trivia question on CompuServe in the mid-80s, listing occurrences of “MZ” in various DOS thingies and asking “Why ‘MZ’?”
“At the start of the project that eventually became Microsoft Exchange, they chose X.400 as their model, seeing as X.400 positioned itself as the industry standard for email interchange. (You can see for yourself how well that worked out.)”
For those who are wondering… The reason X.400 lost was pretty simple – administrative overhead.
An X.400 system has a domain, but not in the DNS sense. The domain is simply a grouping container for all the addresses and hosts in that email system – more akin to a LAN Manager/Windows Domain.
If you want to send emails to users in another domain, then an administrator of your domain needs to add some discovery information about the other domain to their system. Typically, this will be the name of the domain and the address of a host that will accept emails for that domain.
This was accepted as normal.
By contrast, SMTP also has domains. But because SMTP is a product of the internet, it was designed to be distributed. Originally you needed to know the host computer for your recipient’s emails, but after a while the internet moved to the MX record – a DNS entry that basically says “you can deliver emails for users at this DNS domain to (one of) this host(s)”.
With the MX record, SMTP email transport becomes effectively self-configuring. Your email administrator doesn’t need to do anything to send email to any domains unless there’s some tricky routing for security purposes – the email server simply looks up the MX record for the target domain, and transfers the email.
When I started working with email systems in 1995, X.400 was still quite prevalent. The first few links I set up were all X.400. But a mere couple of years later, X.400 was a dead man walking. You only needed to know about it because the big software products – Exchange/Notes – were based on it. Every link you created was pretty much SMTP, occasionally with a relay host, but often just direct to the internet.
The change was really quite sudden and pronounced – email was one of the key tools everyone expected from connecting to the internet, and SMTP made email connectivity scale in a practical way that X.400 could never match.
X.400 was dead before Exchange ever shipped. Speaking as one who was there, when what became Exchange (not including the message store) was designed in the late 1980s, it wasn’t even clear that TCP/IP was going to be the widely adopted commercial solution. Lots of smart money was on the ISO/OSI 7-layer model, and that’s the networking world that Exchange was designed to fit into, supporting both X.400 mail and X.500 directory services. You may laugh at the administrative overhead of X.400, but at the time the TCP/IP world was even worse. As an example, DHCP wasn’t even defined until 1993. Before that if my work computer needed to have a TCP/IP address it came to me in inter-office mail, in the form of a sticker I had to attach to the side of the computer. Addresses got recycled into the available pool by having workers peel the stickers off of old computers being junked.
Anyway, around 1994, with Mosaic invented and the world turning rapidly to TCP/IP, it was clear to us that the OSI model was going to fail in the market. That left us in an awkward position. We could start over on Exchange and build a native “Internet” style mail system, but that would have blown our schedule out of the water and possibly our jobs out of the company. So instead we beefed up the Internet gateway and shipped what was essentially an OSI mail system wearing an Internet disguise, slowly migrating the internals to natively use Internet protocols over the next several releases.
And yes, steveb sent a lot of email, all of it in lower case and very little with punctuation.
1993? I was so surprised I actually double-checked that, and indeed, the DHCP RFC came out in October 1993. Wow. I thought that most of the basic stuff got done in the mid to late 80s. So I guess it wasn’t until the early to mid 90s that we got something that resembles the modern landscape. I guess NAT was even slightly than DHCP later, but I’m not sure if that should count as important.
Dynamic IP address configuration is defined in RFC 951, dated September 1985.
It’s amazing that I use a version of this as late as 1994 (dialup SunOS shell account, mail, news, lynx, talk) before our Internet provider implemented PPP. I even remember using a graphical web browser that worked over shell combining Lynx with zmodem to deliver images. PPP was available by then, but was often broken, while shell access was still available.
Lynx and zmodem? Do you remember what the browser it was called? That is an awesome hack.
NCSA Mosaic
Slipknot
Thanks for bringing back fond memories. I used Slipknot back in college; it was slow but it was infinitely better than using Lynx. If only I had known I should squat on some good domain names while they were available.
“Commenter JT seemed quite astounded that Steve Ballmer sent email.”
Although there may be some younger people who don’t know how long email has existed, he may be referring to the many CEOs and other high level executives who do not use email. For them, anything that involves typing on a keyboard is viewed as a menial, secretarial function (the CEO of the company I work for falls into that category).
Yes, but CEOs need to be extra careful about what they email to the company and how they present themselves, and not all of them come from a writing background. It should’ve become less of a problem after 1995, though.
Fortunately the first e-mail system that I used was post-bang path. It was cool to be able to send stuff to people anywhere, but with the UUCP gateways in place you couldn’t do much more than send text and the occasional (small) uuencoded attachment. If you wanted to talk with most people you hopefully had a MCI or CompuServe account because the gateways were awkward to use and often didn’t work in my recollection.
I recall X.400 being the next big thing, replacing every existing technology with this uber standard. It was big, I’ll grant, but it was so complex that I don’t think it would have ever worked universally; look at how well SGML, TIFF and all the other everything standards fared! At least we got LDAP and other technologies from the work done on X.400.
X.400 was the email system. X.500 was the directory, so you have that to thank for LDAP.
I was, unfortunately, working on creating an X.400/X.500 setup back in 2000.
I can’t remember what was supposed to doom the Internet first – when the amount of material generated daily on Usenet was greater than a 1200 baud modem could continuously transmit or when the map of the entire Internet would no longer be printable and you wouldn’t be able to route your email.
> (Uunet was a major service provider at the time, so everybody knew how to reach uunet, at least. The uw-beaver server was known to uunet, so once your message reached uunet, it would make it the rest of the way to its destination.)
In the early 1980s there were several well-known hub servers like uunet and decvax, but you as the sender were still responsible for defining the full path from where you were. So if you were at a computer named seismo and seismo was connected to uunet you would need to say seismo!uunet!uw-beaver!microsoft!billg. Later they added crude routing so you only needed to know a person’s address from a well-known server.
I believe the Information Superhighway was originally a reference to the Internet as a whole. It’s unfortunate how many people think the World Wide Web is the Internet. Sigh. Anyway, I remember the unattributed quote, “Getting information from the Internet is like trying to take a sip from a fire hydrant.”
Oooh it’s always nice to feel young ;-)
Rather before my time, but that makes it just the more fascinating to read about.
Steve Ballmer still (or did quite recently) send email actually. I think if you read between the lines in the blog too it suggests that he also seemed to happily get involved in things somewhat down from executive-level, and probably quite well respected.
I believe Bryan Willman was the ‘W” in WZMAIL.
Yes, that sounds familiar.
Hey, I got referenced. I could’ve been clearer, of course, but what I actually said was “I believe JT is amazed that a Microsoft executive would send an email in 1992”, meaning I didn’t really think he’d be surprised at the _existence_ of email, but more that it existed in a form suitable for comfortable, everyday use by a Microsoft executive, supplementing or replacing the fax and memo, as opposed to still being a specialized system intended for developers or computer researchers on a university campus. I started being aware of its everyday use around 1995.
“We are not yet committed to supporting RFC 822 in our Internet gateway.” Wow! Even back then dev teams were finding ways to turn away legitimate bug reports.
Fun fact: The first demonstration of the prototype Internet was coordinated using email. Yes, I know!
Until 1982, “internet email” was just another non-interoperable email system: It was a local system (such as has existed since 1965) with (since 1971) a cobbled-together system of inter-machine transfers using a combination of cron, uucp, chewing gum and string.
Prior to 1993 there were multiple competing email standards, including SMTP and X.400, but it was not at all clear that SMTP would win the protocol wars, particularly as until the late 80s commercial organisations weren’t even allowed on the internet.
Yet, businesses used email before they were on the internet. Gateway products allowed exchange of mail between different protocols and systems such as PROFS, MHS, ParaMail, OfficeVision, CC:Mail, MCI Mail and SNADS.
X.400 lost because there were no free implementations, wheras you could do SMTP in Awk.
Freemium is the master marketing model.
—”Well, you can see who won that battle of the standards.”
No, not really.
My best guess would be “Microsoft lost”, seeing as how MAPI/RFC is being dumped for EAP. But, like I said … a guess.
No mention of Ray Tomlinson? ;)
Everyone knows V. A. Shiva Ayyadurai invented EMAIL, mainly because anyone who doesn’t know this gets sued: https://www.techdirt.com/articles/20140907/06302728447/huffington-post-finally-removes-all-articles-about-fake-email-inventor-meanwhile-he-threatens-to-sue-his-critics.shtml
It was invented in 1978, so any references to EMAIL, email, e-mail, or anything similar prior to that are the products of a disturbed mind.
I believe that Dan Lipkie was also a co-writer on WZMail.
A common way of referring to Microsoft addresses from the “backbone” (or Internet) was with this notation:
{uw-beaver,decwrl,fluke,uunet}!microsoft!billg
The bracketed nodes were directly on the backbone. Sometimes you would find the “microsoft” without the final “t” so it would be:
microsof!billg
Later we could use the notation:
microsoft!billg@uu.uunet.net.