Computer history: Wargaming and FEL-LAN (1983 – 1986)


Wargaming and FEL-LAN

Wargames on the move

The Operations Research group (later department) performed Wargames and other demonstrations at remote locations of the Armed Forces using long-distance data communication connections. One of the online and real-time exercises involved linking a defence communication truck somewhere on the Veluwe with the Cyber mainframe in The Hague via some military microwave links. PTT towers were used as relays. The problem was the generator of the communication truck. The frequency gradually went down from 50 Hz to 45 Hz. When reaching the last frequency, the signals produced by the terminals were too slow to be recognised at the Laboratory side of the link in The Hague. Manually adjusting the truck’s generator frequency every twenty minutes solved the problem.

Another Wargame exercise linked the airbase Twente to the laboratory. A Royal Netherlands Army truck with a high antenna mast with a microwave transmitter on top was positioned in the courtyard of TNO-FEL on Waalsdorp. Twelve terminal lines entered the laboratory through a window. During the preparation week, the multi-hop connection was only up and running for a few minutes as the conscripts at intermediate masts liked to listen with their radio equipment to normal radio stations. The solution was found by a ‘high-ranking officer’ on Friday morning: if the connection was not really operational by 1 pm, all conscripts’ weekend leave passes would be withdrawn … within one hour the whole link across the nation was stable and flawless.

To prevent these problems during the next Wargame editions, a MicroVax was purchased on which the Wargame program would run. Preparations for the next demonstration at the Twente Air Base were disastrous: the magnetic disc slowly degraded and became unreliable. After transport to another room at the airbase and after a full reload of the system there were no problems. Back on the demo stand, the disk started failing again.
Call: “Did we have an idea at FEL-TNO in The Hague?
What kind of room are you in? Is it may be too hot?
We look out on the tarmac; a very large situation briefing room, not too hot …
Uh, outside the window is a rotating radar … that does not hurt, isn’t it?
Remembering our previous problems with radiation by radars on magnetic tape units, we suggested shielding the system from the ground radar somewhat by turning the system 90 degrees. The observed error rate decreased. We asked the colleagues to note the physical dimensions of the system. One of the mechanic engineers at the laboratory was sent out to buy chicken wire. Within a few hours, a Faraday cage for the MicroVax was built that was moved and installed at the airbase. The next day, the chicken-wired MicroVAX was less prominently placed on stage in the demo room of the Twente airbase but worked flawlessly.

Apart from the Wargames, the Operations Research division taught courses twice a year at the Hogere Krijgsschool (HKS). The Soltau courses were supported by a computer program which ran on the CYBER. It was always a burden to free enough terminals for those events.

Local Area Network study and lab-wide LAN installation

There were 480 terminal lines between the laboratory rooms and a huge patch panel. In addition, there were 21 external PTT leased lines ending on a separate patch panel that could be switched off with one switch for security reasons. All those incoming lines could be connected with 217 lines to the computer systems: 128 to the Cyber, 40 to the PDPs and 49 to the VAXes. Some were hardwired patched; others connected to the DATUS port selector. The port selector was a kind of telephone exchange for data communication links.
The DATUS-port selector configuration was expanded many times, but the exploding demand for terminal lines could not be satisfied anymore. As users many times found all ports to the CYBER being occupied, a second 2551 with eight CLAs (32 asynchronous lines) had to be installed.

In 1985/86, the lab’s IT policy committee started a project team that had to investigate how TNO-FEL could install a Local Area Network (LAN). That included researching the security aspects of a LAN. A student of the Technical University Delft gave support to that work. He had to look at the security aspects both from a theoretical and practical view.
This practical research, therefore, included a mini-network based on the DECnet which connected the three VAX 11/750 (VA, VC and VG) systems. A PDP 11/34, a PDP 11/44 and a PDP 11/725 were also connected to that network. As a good researcher ‘hacking’ (group professor Herzberg), the student was able to run programs on the VAX for administrative purposes (VA) without having any authorisation to use that system. On the other hand, LAN technology promised a lot: a throughput of 150 MB/hour was measured on the VAX 11/750.

After a thorough study, the workgroup asked several potential vendors for a network design and a price quote. LANs at that time, especially when the LAN had to support equipment of diverse manufacturers, were a ‘terra nova’ in The Netherlands. Well-known computer companies foresaw ‘earthing problems’ and excluded such errors from their guarantee. Obviously, they did not understand the IEEE 802.3 network standard that was asked for by the Laboratory. Otherwise, they would have understood that such earthing problems could not occur in a properly designed and installed network. In the end, the committee advised commissioning the installation of the network to Control Data Netherlands BV. Their proposal included subcontractors for the installation itself under their responsibility. The FEL local area network, the FELLAN, was installed as a star-shaped network around the central data communication room. It consisted of ten thick Ethernet segments of 500 meters each.
The LAN transceivers were delivered by another company. The FEL was the only organisation in the Netherlands that thoroughly tested each transceiver. Transceivers were only accepted when they passed the IEEE 802.3 specification tests. Even marginal differences in the specifications were not accepted, despite the fact that those transceivers worked in other networks flawlessly.
The DECnet used a pre-IEEE 802.3 standard based on the (Xerox) XNS-header specifications. In principle, the DECnet data layer specifications were in conflict with the IEEE 802.3 international standard. In practice, however, the length field did not conflict. Thus, both the systems that met the IEEE 802.3 specifications and DECnet-based systems could simultaneously use the same Local Area Network (LAN), although the network managers had to take care of the right configuration of the transceivers (with/without “heartbeat”).

In 1986, the Laboratory installed a Datanet-1 (DN-1) connection – an X.25-based communication public network link – making it possible to reach external computer services as well as bulletin boards of the computer suppliers.