12

With the advent of microcomputers, there was a large number of manufacturers entering the home computer market. The ability to network these computers together wouldn't be very useful (or affordable) for home owners, but businesses and education/research establishments may have multiple computers, and a desire to create a local area network.

To this end, which manufacturers designed successful, bespoke networking systems for the microcomputers they were making? A brief description of the systems and their merits would be appreciated.

I'd consider a successful system to be one that was supported on several different machines made by the manufacturer (and possibly other manufacturers' machines) over a period of time.

I'm looking mainly at the 8-bit era computer systems that were aimed at the home/small business market. I'm not looking for networking systems that were also developed for minicomputers and mainframes (such as IBM's Token Ring), just those designed for micros.

closed as too broad by pipe, Chenmunka Feb 7 at 18:11

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 1
    @SolarMike perhaps using the ST’s MIDI connectors, at least MIDI Maze could do that. I’m not sure that qualifies as a full-blown network ;-). – Stephen Kitt Feb 6 at 15:29
  • 1
    This is a list question and, as such, may be closed as too broad. The simple answer is "most of them". – Chenmunka Feb 6 at 15:33
  • 1
    A wide topic, but a really interesting one. – Mark Williams Feb 6 at 15:36
  • 1
    The exclusion of IBM is unwarranted; they had a low-cost LAN network too. It's arguably the most long lived as its NetBIOS lives on in Windows. – user71659 Feb 6 at 20:09
  • 2
    @MauryMarkowitz Can you find another site without that text? I don't know why you think why retrocomputing is so special that it can't work within the parameters that made the rest of stack exchange into such an awesome place. I strongly disagree with your position. – pipe Feb 7 at 13:09

10 Answers 10

13

Preface: As with many early questions, it's up to the definition of network. It might be helpful not to tie this too close to our modern understanding of a connection between (mostly) equal peers.


Well known networks on the micro/home computer side may be of course

  • Acorns Econet introduced with the Atom in 1981
  • Sinclair's ZX-Net introduced with the Spectrum Interface 1 in 1983 (also used for QL computers)
  • Apples Apple-Talk introduced in 1985 for use with Mac XL (Lisa) and standard Macs, as well as the IIgs

But then there where literally hundreds of solutions of early networking systems centered around shared use of data storage and other resources, like printers.

There were some repurposing concepts of existing networkings/busses

  • Already for the very first PET computers (as soon as there was a disk drive) third-party drivers enabled (*1) the use of multiple 'controllers', so several PETs could share one (or more) drives (*2).
    • In a most primitive way, data exchange could and was done by using the drives as 'mailboxes'. Beside a very small driver to enable multiple controllers, no additional software was needed. Application software used regular disk I/O to handle this.
    • With a little more software, monitoring the bus while inactive, direct PET to PET communication could be done. Again using regular I/O commands.

As well as others, made special to type. More often than not based on serial connections. Such systems where often used in classrooms as well. This was already driven in 1980 as far as having micro based 'terminals' connected to a central storage server system. For example Intertec's (*3) CompuStar setup. Quite like way later work station setups.

-

And then there are of course network solutions independent of manufacturers - like using amateur radio to network ... but that for sure leaves the LAN category.


*1 - A controller in context of HP's HP-IB/GPIB (or IEEE-488 as called when standardized) is a device that tells others what to do - like sending data or receiving. Usually a computer. By definition the standard does allow the operation of multiple controllers (thus computers) in one network, but Commodore did not implement that part. So it was up to grabs for others to provide an extension.

*2 - I hope this isn't disqualified, even though the GPIB was handed down from professional equipment (including mini computers) :))

*3 - More recognized today for their 1979 Twin Z80 Superbrain system - which, without disks also happened to be the workstations (here called terminals)

  • GPIB may have been handed down from minicomputers, but it's certainly worth mentioning. The ZX-NET is totally new to me, I'd love to read more about it. – Kaz Feb 6 at 18:56
  • The Atari 8-bits had a network, ~96kbps IIRC, built by a Michigan educational consortium. Basically the same idea as the PET system, for sharing drives and printers. – Maury Markowitz Feb 6 at 19:52
  • @MauryMarkowitz Got any link for that? – Raffzahn Feb 6 at 20:34
  • @Kaz There are many more, especially for the second list, but I need to check some old papers for that - gimme some more time. – Raffzahn Feb 6 at 20:36
  • 1
    The simplified (and cheaper) hardware spec of localtalk, kept Appletalk limping on for some time (more or less until ethernet prices began to come down). But that's a bit latter than the question addresses. – dmckee Feb 7 at 5:57
23

One prime example is the Econet networking system designed by Acorn Computers.

Best known for their BBC Microcomputer and Archimedes systems, Acorn started with a range of modular rack-based expandable computers known as the System range, released from 1979 onwards. These machines were aimed at serious hobbyists, researchers, and industrial users, and could be heavily customised with a range of expansion cards, including Econet network cards.

Econet uses a bus topology, with a four-wire backbone carrying differential signals. One pair of wires is used for a network clock signal (generated by one node, or a dedicated clock box), with the other used for data transmission. The clock speed, which controls the data transfer rate, can be customised to suit each network installation: shorter networks can support a faster clock, but the clock must be slow enough to accommodate the slowest computer on the network.

Up to 255 nodes can be connected to an Econet, and multiple Econets can be bridged together with gateway nodes. (The individual Econets can be running at different speeds.) A node can act as a file server for other nodes (originally using floppy disks), or as a print server. File Server software was released in various "Level"s of complexity, which added support for user passwords, subdirectories, and hard disk storage among other features.

Econet hardware was developed for the BBC Micro and Archimedes machines, and as a result many schools were fitted with Econet networks. In later years, with the growing adoption of Ethernet, Acorn developed Acorn Universal Networking, which added support for Ethernet and TCP/IP, while maintaining backward compatibility with Econet protocols.

More information, and copies of original documentation, is available here.

  • Actually, the BBC Micro was designed and marketed primarily as an educational machine, not just for schools, for everybody. Anyway, schools probably were the biggest adopters of Econet. – JeremyP Feb 7 at 9:15
  • I thought I had posted this earlier (erased? forgot to press Add?) but there is a claim by Herman Hauser quoted in Wired that Cambridge Ring (circa 1974) was the basis for both Econet and then AppleTalk. This is not true; the original claim is based on the statement that Apple used the same comms chip as the BBC, but one can trivially verify such is not the case. – Maury Markowitz Feb 7 at 12:57
6

You can also consider GPIB/IEEE-488 to be a network. It's a "bus" but I think at this level the distinction is debatable. It allowed multiple devices, they were connected via daisy chained cables. You could either broadcast or send packets to individual devices. While there was a dedicated controller, devices could request access to send data (so it wasn't master-slave per se, although it was mostly used that way).

Typically it was a "smart host" with "less smart" devices (i.e. a computer and assorted peripherals) vs a computer to computer interchange. But it could be used as one, just having one computer configured as the controller and the other as one of the devices.

It started in the late 60's, became more popular during the 70's and 80's, notably with Hewlett Packard test equipment. It's also notable that you could use the interface with the HP-41 calculators, and it was also built in to early Commodore computers, such at the PET. A version of if exists in the C64.

It didn't follow a formal networking OSI model, it came before that.

But I think it certainly qualifies as an early network.

  • 1
    "It allowed multiple devices": Why do you use the past tense? The IEEE-488 bus is still in use nowadays in many laboratories. – Massimo Ortolano Feb 6 at 20:43
  • There could be multiple controllers as well - on GPIB at least, commodore didn't implement that part. – Raffzahn Feb 7 at 10:58
6

In 1984-86, I installed several Digital Microsystems (DMS) HiNet networks in the UK. This was an RS422 bus system running at 512 kb/s in a master/slave polling configuration.

The original server had an 8" hard drive and an 8" floppy drive, running a CP/M-based operating system. A later version was 8088-based with 5 1/4" disk and diskette.

The workstations were diskless running CP/M 2.2. There was also, later, a dual Z80 / 8088 workstation and a Z80 PC card.

I just found a link here :http://www.retrotechnology.com/dri/digital_systems_products.html

Customers I dealt with included Distillers' Company and the Audit Commission. I understand that Gatwick Airport also used DMS for their flight information displays.

6

Sinclair's ZX network was implemented in both the ZX Spectrum (with Interface 1) and the Sinclair QL - Some QL compatibles that came later implemented it as well.

It was a serial multidrop network with max. 64 devices on the line and advertised 100kbps transfer rate (measurements show it actually uses a bit rate of 87.5 kbps) that was bit-banged by the CPU to a two-wire simple audio cable.

Some "carrier sense" was implemented in the potential sender by listening for network activity for some ms before actually occupying the network with a so-called "scout" that contained a break plus the sender's address - if that could correctly be read back from the line, the sender assumed to own it until the next collision.

The ZX Spectrum actually only got minimum messaging channel based transfer (that needed active collaboration between the clients), while the QL could, with additional software, implement a file server. You could access all device drivers of the remote QL running the file server, even open screen windows, and printer channels. The QL could talk to ZX Spectrums and vice versa.

The Elan Enterprise had a similar serial network built into the basic box with similar features but used four wires.

6

Research Machines in the UK had the Z-NET network which was used with their 380Z, 480Z, Nimbus 186 and Nimbus AX/VX computers, starting in the early 80s. It was CSMA/CD over coaxial cable/BNC at 800kbps, later 2Mbps, using Microsoft's MS-NET operating system to run a fileserver.

This was based on the Zilog Z8530 SCC - it appears to be a product of Zilog (see brochure, paper at SIGSMALL'80, YouTube demo) so it's technically out of scope for this question, but I don't know any other computers that shipped with it.

5

IQ 151's could be connected by their own proprietary network (VARIEL). The server was running CP/M, and could share floppy drive, printer, transfer BASIC programs to/from clients, in later versions and with the GRAFIK expansion card also read clients' screen. In later versions, the network could be joined by PMD 85 as well.

There is an online user manual available (in Czech).

5

Although IBM had the mainframe-oriented SNA and later Token Ring, which was posed as a competitor to Ethernet, they did have a small office-scale LAN solution, the 2 Mbps IBM PC Network.

This was released in 1984 and was intended to support IBM's PC line. It apparently made it into the PS/2 Micro Channel era.

Arguably, this was the longest lived solution because it introduced NetBIOS, whose descendants live on today in Windows networking.

3

Between 1982 and 1985, Thomson and others developed the Nanoréseau that was used to link several TO7, TO7/70 or MO5 machines together.

These setups were mainly deployed in french schools as a way for teachers to monitor and install software for students.

There is a whole website dedicated to its legacy (in french)

  • 1
    Nanoréseau wasn’t developed by Thomson. – Stephen Kitt Feb 7 at 9:10
  • From the linked article, it seems Thompson joined the Nanoréseau project in 1984, two years after it had begun. – Kaz Feb 8 at 11:26
  • @StephenKitt I edited my post – Wizou Feb 8 at 11:59
  • 1
    Thanks, but your answer still gives the impression that Thomson was the major driver behind Nanoréseau’s development, when it really played a rather minor part at the time, with most of the development being driven by USTL and Léanord. I don’t think it’s fair to claim that this is a LAN system developed by the manufacturer of the micro-computer it was used with, which is what the question is about. – Stephen Kitt Feb 8 at 14:21
0

Regarding networking protocols, X.25 is one of the earliest. This was used by various commercial providers of networking and time-sharing systems such as CompuServe, as well as systems such as the UK's JANET which linked educational establishments.

In any discussion of early networking, we should also mention ARPANET which linked various educational establishments in the US, and eventually evolved into the Internet we know today!

  • 1
    Those are standard (ish) networking protocols; that seems contrary to the question which asks "which manufacturers designed bespoke networking systems for the microcomputers they were making?" (my emphasis) – Toby Speight Feb 6 at 17:42
  • Agreed. X.25 and ARPANET would also be larger than "Local Area Networks" (see question title). – Kaz Feb 6 at 17:55

Not the answer you're looking for? Browse other questions tagged or ask your own question.