26

Although ARPANET was invented in 1969, the Internet as publicly available infrastructure didn't really become available until 1989. But people were certainly using computers as communication tools in the 80s.

For individuals with modems, there were local bulletin boards, and large dial-up services like CompuServe and Prestel.

But what about corporations? Take a multinational corporation like Exxon, that was certainly no stranger to either computers (payroll on IBM mainframes, technical calculations on supercomputers and workstations, and increasingly large numbers of IBM PCs and compatibles for running e.g. Lotus 1-2-3) or electronic communications (the chicken and egg problem for fax machines was solved by such corporations initially buying the machines in pairs, for offices in different cities to communicate with each other). It would certainly have been useful to have email between workers at all distance scales from within a building, to between continents.

What solutions did they typically use for this?

13
  • 25
    Leased phone lines (which powered ARPANET as well).
    – Jon Custer
    Feb 13 at 14:04
  • 7
    X.25 packet-switched networks were offered by various companies for both terminal access and computer-to-computer use.
    – HABO
    Feb 13 at 14:49
  • 38
    "Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway" - anon Feb 13 at 15:24
  • 8
    @GiacomoCatenazzi - I was sending (text-only) email from my university account to my brother's university account (at another university) by 1982, using Bitnet to Arpanet gateways.
    – Jon Custer
    Feb 13 at 17:11
  • 18
    @GiacomoCatenazzi "Forget email and such things. Computers were not connected." Uh, yes they were. The question is about the 1980s. DECnet dates from 1974. DEC had a world-wide network for its own engineering use from maybe the late 1970s. It reached my UK-based 11/70 in 1979 or 1980.
    – dave
    Feb 13 at 17:44

6 Answers 6

39

Corporations needing such connectivity (which was still unusual in the 80s, outside the realm of large corporations) typically used leased lines, i.e. private point-to-point connections leased from a carrier. T1 and E1 were available in the late 70s. These could be used to connect mainframes to terminals (e.g. over IBM SNA or DECnet), and later on to build Frame Relay or ATM networks, or even connect to ISPs’ points of presence for general Internet connectivity. Large-scale networks, including ARPANET, were themselves built on top of leased lines.

Leased lines (typically dark fibres now) are still in use today.

In the 80s and 90s there were quite a few different networks, many academic, that some companies were able to participate in; for example BITNET, EUnet etc. Email between companies often ended up using services like CompuServe or CIX. Bear in mind too that reading magazines from the era gives a skewed perspective: IT journalists were far more likely to have access to some form of email than people even in large corporations. In the 80s, even a LAN was unusual, and sneakernet dominated.

Various other technologies were used, e.g. private microwave connections between distant sites with line-of-sight.

7
  • 2
    Also leased comsat channels.
    – John Doty
    Feb 13 at 14:47
  • 6
    @MarkRansom I’ve always heard leased fibres referred to as dark — they can be leased because they’re dark. (Obviously they’re no longer dark once they start being used.) Feb 14 at 16:20
  • 7
    AIUI when it comes to buying connectivity "dark fiber" means the fiber's owner only provides the fiber, and it's up to you to "light" it. This contrasts with services where the fiber's owner lights the fiber and sells you a service at a higher level. Feb 14 at 19:37
  • 3
    Afaict a major factor is distance. If you buy a link from one side of town to the other, then it is quite likely to be dark fiber. If you buy a link from the US to the EU it will NOT be dark fibre. Feb 14 at 19:40
  • 4
    A dark fibre is dark because to the end user, it comes out of the wall unpowered and unlit. The customer is expected to fit their own optics and is buying a "long piece of fibre". Compared to any provisioned service which is normally presented as RJ45 ethernet, and may be transported in a VLAN or similar in the middle where it is invisible to the end user.
    – Criggie
    Feb 15 at 1:47
27

If you had the need for WAN in the 1980s (not many, even large corporations did, IT was still often built as islands, maybe with the exception of banking, finance and accounting), you had the following options (Basically in increasing order of fees). Not all of these options (except maybe the satlink) were available globally:

Analog leased lines - That's practically a permanent two-wire phone line you rented from a carrier. You had to use telephone modems on either end (so "one-time-dial-up", the connection was never terminated), and obviously the available bandwidth was very limited, typically 1.2kbps or below. Better modems became only available in the 1990s.

Digital leased lines - T1 (1.5Mbps, in the US) and E1 (2Mbps, basically everywhere else, and highter bandwidths in the telco digital hierarchy, up to 155Mbps were available, if you could afford them), that's a synchronous serial connection typically ending in X.21. Needs data, some control and a clock line. Here, the telco provided you nearly direct access to their internal PDH/SDH/SONET network, or in some cases, if practical and feasible, they might even have built a dedicated microwave connection for you.

Dark Fiber - This was a dedicated, physical fiber connection between your endpoints that didn't really use the carrier's network (only that single fiber and their right-of-way), but was rather handled like a piece of wire by them. With more restrictive carriers, that was the only option where you could (and had to) provide your own terminal equipment. So, in essence, what you pushed through that fiber was basically up to you and your equipment.

Satellite link - You could, after commercial sat TV was established, rent a transponder slot on a geostationary TV satellite (basically one TV channel), which gave you in essence an E1 (2Mbps) like a digital leased line, but with an abysmal round-trip delay. This was indeed specifically used by multi-national airlines that, in some countries, didn't have any other option to reliably reach their local premises. The physical presence after the satellite modem used to be X.21 as well. Some national telcos with special needs (e.g. Germany after the re-unification that was faced with a huge demand, but only a rudimentary network in the former GDR) even offered packet-switched X.25 (see below) over satellite and built some sizeable satellite-based networks.

Packet-switched network (X.25) - Some national telcos ran local packet-switched X.25 networks you could join (Datex-P in Germany, DATAPAC in Canada, numerous in the US, for example). The appearance was close to what we know as TCP/IP networks today, but the internal working was much more complex. These provided packet-switched endpoints (on physical X.21 interfaces) with typically 64kbps of bandwidth. Although the networks as such were packet-switched, in practice they mostly appeared like leased lines to the end user (because the endpoints were configured to connect to a fixed other endpoint - there was, even if the networks could have provided that, no interconnect between random subscribers). Banking and Finance were typical customers.

With regards to "how common were such WAN installments?" it was like it was today: Anything a business spends money on must have a business case and an ROI, and WAN came attached with sometimes very significant cost to build and maintain. Most producing businesses "worked" and were organised in a way that allowed them to work and only if you could achieve significant advantages you made the investment. Remember, in those days you could assume a (paper) letter to reach its destination the next day and the postman delivered twice or even more times a day. So, in some industries, takeup was slow, in others a lot faster.

17
  • 4
    "(not many, even large corporations did, IT was still built as islands)" ? What makes you think that? In the 1980s state and national networks were a quite common sight. Maybe not for people on the street, but anyone in IT. Joe Average only noticed the new ATMs with up to the minute booking and his insurance company having the data at hand. Sure, they were big corporation, banks, insurance, car makers and alike, and so where their networks. When I returned (1981) from service our department had a network of 3 mainframes and nodes spanning southern Bavaria, ~300 km East to West.
    – Raffzahn
    Feb 13 at 17:50
  • 3
    Re: satellite links. The DEC engineering network USA-UK link was from a very large satellite disk in the backyard of DEC's Littleton MA site, to the GPO's Goonhilly Downs site in the UK, and from there via post office leased line to DEC Reading UK. I forget the speed, though I do recall that satellite latency required protocol changes to allow larger ack window sizes.
    – dave
    Feb 13 at 17:58
  • 6
    @raffzahn What makes me think that? Well, personal experience. I worked for a large multi-national company (>100000 employees) in engineering and production (not finance or insurance, which might have been different) at that time, and until about the mid-80s we sent around couriers with tapes full of CAD data every night. No networking. I got one of the first five(!) e-mail addresses in that company in, if I remember right, 1986. Yes, IT was organized as islands, especially in remote sites (ours was)
    – tofro
    Feb 13 at 20:16
  • 5
    @Raffzahn To sum it up: You were in a lucky situation, I wasn't ;) But I still consider what we had at that time as "normal", and not outdated for the time (Still agree that banking, accounting [partial, if you had to] and insurance may have been different). BTW: At time, the accounting and payroll data tapes were on the same car as the engineering ones...
    – tofro
    Feb 13 at 20:39
  • 3
    X.25 networks weren't just from national telcos. There were commercial providers like TELENET and TYMNET in the US. Compuserve started as one of these, then they created the consumer product so the network wouldn't be idle outside office hours.
    – Barmar
    Feb 14 at 15:01
15

Leased Lines

Leased lines are always at the bottom of each network, no matter whether they are made as twisted pair of galvanic coupled wire, dynamic amplified lines, fiber or satellite link. They may be switched or not and likewise independent of end point equipment (modem) being included (telco supplied) or not.


History Repeats

It may be a surprise, but public shared/switched lines are a newer development than private/leased lines. Even back in the age of telegraph and Morse code, companies built or leased lines for private use. This continues with teletypes and phone lines further on. After all, it was always cheaper for large corporations to connect their subsidiaries via a leased line than to pay for each phone call or telex.

Switched phone network only came after private and leased lines.

And as with earlier teleprinters and phones, private lines were built or leased for data communication. Installing data networks atop those lines goes back to the late 1950s/early 1960s - including switching - and was pretty standard in the 1970s. Of course, this was only appropriate for large(r) companies and organisations - as before with phone and telex. Dial-in was only a minority use case - and even those were usually only using the next local office.

At some point during the 1970s telcos started to offer higher level products than just leased lines, including high speed modems and virtual connections, generating additional business by taking off low-level duties from company installations.

So essentially everything we now associate with (internet-like) networks was there, long before any public offerings were made. And long before telcos saw a business model in offering communication networks to a wider audience instead of only leasing lines. They repeated the earlier development during the 1980s with small and mid sized businesses in mind (*1) - doing the same step as half a century before when they created switched phone access.

And the same thing repeated during the 1990s when companies like AOL built their own networks on top of leased lines - just this time for the general public. And like before with telex and phone and later data switching, telcos saw the market and added/opened their network directly to customers, eliminating the middle man, aka AOL&Co.

Bottom "line": Nothing new since the 1880s, just repeated cycles of broader availability, all built on the same base structure.


Size May Vary

To be clear,this is about use in mid-sized to large corporations, as asked in the question - not some 5 man store with a mini in some corner - also, it's from a time before PCs or micros in general became a thing, ~1980..1985.

As an example of a smaller installation in the early 1980s, we can look at the department I worked for when returning from service in 1981. It was a local service department of a 'reasonably' large hardware manufacturer. Our area was southern Bavaria. For our own business we used two mainframes - one production, one for development - and a backup in Munich. Another installation was in Augsburg, a city about 80 km west of Munich, connected with a leased line running at 19200 Bd. Other connections existed to several local offices around the state, all leased lines running mostly at 9600 Bd. A network spanning about 300 km East to West. Maybe relevant here, we had no real IT budget, so our hardware was over all some 5+ years old ... about the age when our customers dumped and updated theirs, so we usually just rerouted what otherwise had been scrapped into our basement (literally a part of the parking level :)).

Our customers in turn were banks, insurance companies, even more banks, a not-to-be-named car manufacturer and so on. They had state- and nation-wide networks which were way bigger than that little mentioned. When in the 1970s bank branch offices still were running batch services using dial-up at night, by 1980 most of their branches were connected with leased lines, allowing direct access to their mainframe systems - likewise operating of the back then spreading ATMs (*2). Even small banks with only statewide operations had a hundred or more nodes connected with leased lines.

Larger nationwide banks had 1000+ branches and multiple computing centers, all connected by leased lines - with speed going into megabit ranges, including early fiber and satellite links.

In 1983 our little 'private' network and its (two surviving) mainframes was finally allowed to connect to the company's main network. I mentioned already it being a 'sized' company. From then on we had true global connections, and all of that on leased lines all around the globe, as there were computing centers or at least communication nodes literally in 100+ countries. All predating the Internet, all without IP and at most a few PCs peeking in, camouflaged as terminals :)

Take Away #2: For larger and international companies data networks were standard and in wide use.

The latter is also a reason why the switch to IP and generic internet infrastructure did at some places drag until the 2010s. Why dump a working infrastructure and in turn acquire all the issues of an open network?


P.S.: At that point it's often mentioned that a boot full of tapes beats a T1 line, even on long distance. All true, but as so often, it's not so much maximum speed but throughput. Anyone who had the job to bring a tape with patches to Russia - or equally worse the US - will tell you that travel time itself doesn't matter compared with all the border (Russia) and customs (US/Europe *3) hassles. That in mind, sending a Megabyte offer a 9600 line becomes very fast and attractive.


*1 - Plus the usual amount of hackers ... err ... technophile early adopters - which they may not as much have had in mind :))

*2 - Nowadays 80% or more of those branches are gone, the ATM being the only leftover.

*3 - It's easy to forget what a mess of borders and regulations Europe was before the EU.

3

To give an example from before the 1980s. In 1974 I started work for a company which had depots and warehouses throughout the UK. Customers phoned orders to their local warehouse where they were entered into IBM 3270 terminals as they spoke. The warehouses were connected by leased lines to an IBM 370/145. At that time the nationalised telecom provider only had 1200bps modems. It became possible later to get 2400bps modems from private suppliers. IBM BSC protocol was used. After a cut-off time, the orders were processed and instuctions printed off in the warehouses to assemble the orders to be delivered to the customers the following morning.

2
  • 1
    My Post Office Handbook of Data Communications says that Datel 2400 service was available in 1968 on 'private circuits', which I think meant leased lines (not the switched telephone lines network).
    – dave
    Feb 16 at 1:01
  • 1
    This Datel brochure implies the P.O. supplies the modems. It's claimed to be from 1976, and the P.O. name rather than B.T. says it comes from the 1970s.
    – dave
    Feb 16 at 1:11
2

Was a short time intern at my local bank in Sweden in 1982 - the local branch computer was connected to the main frame in Stockholm via leased line, so a specialized version of RJE.

0

In 1978 Xerox contemplated a remote document delivery service via a combination of satellite and microwave communications. In 1983 Fedex contemplated a similar service. Neither was built.

1
  • 2
    If neither was built, it does not answer the question "What did corporations use ..."
    – Leo B.
    Feb 15 at 17:23

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .