Welcome to the webpage dedicated
to the H(igh) L(evel) S(ystem).

(please wait a moment for the pictures to be loaded)

This page is dedicated to the HLS. This machine has been the core of the SITA network for a long time but that is history now...
The page can roughly be divided in two sections. The first one is with the photo's of the ceremonial switchoff of the machine, the second is a general description of the hardware and the way it worked. Most pictures are quite small when you look at them but if you click on them you will get a larger version. If you want any of them in a higher resolution let me know, I have high resolution versions of them available.

For any questions or comments please do not hesitate to contact me.
Dries Bessels (Dries.Bessels@SITA.INT)

May 12th, 2000 was the day the HLS was then finally switched off. It has done it's work with remarkable stability and worked hard for the money. Some people tried to revive old skill by for a last time trying to load a tape (not always with a lot of success):

The bottle went round one last time

Then the HLCPLT was ceremonously wrapped around the system units

The country manager (Rob Dickhoff) said a formal farewell to the machine

Jan van Laer (one timer programmer of the HLS) switched the machine to HALT

And Martin Spits (the one person in the world who still understands the hardware) switched the first power supply off

Bert Koster (former operator/supervisor) switched off the second powersupply

And Jan van Laer the last one

One last time

There are a number of people who, when they heard the HLS was being switched off, did not believe it. This has been told many times and everybody heard the comments that Type-B is dead, the HLS is old and defunct, etc. Well, Type-B has a long time to go and is by far not dead but it's functions have been taken over by the MSS in Atlanta. To prove that the HLS is this time really gone:

The history of the machine

For many years the SITA network was based on a by now hugely outdated system called the HLS or to the people who worked with the machines as 'The Philips'. A complete name is a Philips DS714 or DS790 with added to that a Mark I, II, III , IV to indicate which specific version this one was.

Somewhere in the sixties it was decided that the switching of the traffic at that time (all telex traffic) had to be automated and eventually this resulted in the design and building of the Philips DS714 which was at that time definitely the most advanced system ever built. With a number of these machines the first real packet switched network was built by SITA called the HLN (High Level Network). There were typical speeds in this network of a whopping 2400 (!!!) bauds on international circuits.

Late 1980's this DS714 was decommissioned and a new version came in, the DS790. This was basically the same machine but built with more modern materials and technology. Yet, the original program of the DS714 could run on the DS790 with only minor modifications.

The DS790 has worked for over 10 years and one of these machine is currently still running in Singapore. In may 2000 it was decided to switch off the Philips in Amsterdam and we decided to put some commemorative sita on the web for it.

The machine is way to big to be kept as a souvenir or so but I am happy to say that large parts of it will go directly from the computer floor into a to be started museum so this magnificent piece of computer (and SITA!) history will not be lost completely.

The working of the system

The whole idea of the system was that people can send (telex) messages around the world and be absolutely sure that these messages would be delivered. For this a secure 'store-and- forward' system was needed. To ensure that the HLS would store every single message several times. The first storage was on a tape unit and a drum as incoming message. Then it would be processed and after that stored again as outgoing message on tape. The tapes were kept for at least 7 days so everybody could get them repeated in case something went wrong with the (primitive) telex machines. The storage was only there to enable customers to ask for an investigation in case a message was not delivered, delayed or there was something strange going on like fraude with the addressing.

The machine was built up of a processor with a number of control and expansion systems around it. For 'internal' storage there were two separate systems, the 'drum' and seven tape drives. For external storage there were a number of PC's (the retrievals) with SITA's own software running on them.
Connected to the machine were also MIOM's and MXT's which can be compared to the serial ports of a modern PC. As with any piece of SITA equipment it was doubled so for backup purposes there was always an online and a standby.

The original DS714 did not have anything like a console or such. The controls were two fold: One an old style Teletype and the other the front panel. The teletype could be compared to a very old fashioned electronic typewriter, the front panel looked like:

Click here for a detail view of the left side of the panel
Click here for a detail view of the right side of the panel
Unfortunately I could not find an old Teletype machine to take a picture from.

The 'new' DS790 was a revoluntionairy machine in that the teletype and the frontpanel were no longer there, they had been replaced by a Hewlett&Packard terminal! On the terminal you could see a representation of the frontpanel and the keyboard acted like the old teletype system. The screen looked like this:

We will not quickly forget the whole Y2K issue which started to come up 'late last century'. The new DS790 of course was to be certified for this also and passed the test without any problems:

The console was used for programming, starting and stopping the machine. For the day to day work the operators had a range of PC based control systems. The software for this was also known as the PP series. The reason for the abbreviation PP was never completely clear but it was rumoured this had to do with the fact that no budget was available for it so the software was written as a 'Pirate Project'.

First of all there was the 'system' position (PP01):

Note the difference in the two screens. Whenever a certain item changed it would go grey so the operator had a constant and instant view of what was going on.
The screen was divided into 9 windows:
Window 1 - Status of first system - Latest startup time, Software revision, type of startup
Window 2 - Status of second system - Latest startup time, Software revision, type of startup
Window 3 - Status of the subsystems like printers.
Window 4 - Status of the MIOM's and the MXT
Window 5 - Status of the 7 tape drives
Window 6 - The oldest two messages in the system. As far as I know it was only one single time that the system the status 'No messages in system'.
Window 7 - Status of the internal counters of the processor
Window 8 - Status of the retrieval machines
Window 9 - Status of the line monitors.

There were a number of other PC based system connected to the DS790. They were:

PP02 - Low Level drain:

PP03 - Intercept position:

PP04 - Info position:

RPC - Retrieval on PC:

Besides these PC based control systems the HLS was the host system for a new range of products called SITATEX.
- The first version of Sitatex came out in august 1984. It was a program for a monochrome screen, connected asynchronously to the HLS in ASCII and would allow message exchange and document transfers of up to 10 Kb(!)
- Sitatex V2 came out in november 1987 and was the first version supporting colour mode. It was even sold to a number of customers and was conected to the HLS either in Synchronous (P1024C) or asynchronous mode in both ASCII and Baudot.

If the MIOM went down there was a separate startup procedure which had to be followed. This was done with a special 'keyboard' with it's own paper tape reader:

The machine had two cables running to a special card which could be inserted into the MIOM. Then the MIOM could be activated and the program on the papertape could be loaded.

Looking at the processor cabinets you would not see very much:

However, inside these 4 cabinets a lot of things were going on.

There are a few main items here like: The real time clock (bottom left)

The drums (short term storage, top of middle cabinet)

The main processors (middle left and right cabinet).

In the back of the first cabinet were the batteries for the Non-Volatile memory. This was not your average button cell, these were serious batteries:

The MIOM's can be compared to the serial ports of a current PC. Basically, when looked at them from the outside they were just a large row of grey cabinets:

However, on the inside these cabinets were filled and filled with cards. Compared to current computer equipment these are now extremely obsolete
A 16K (!) memory card looks like this, note the little chip in the foreground which has approximately the same storage capacity:

When we open the card you can see the enormous amount of electronic components needed to control such a relatively small amount of memory:

Note the left picture, the small grey square in the bottom right is the actual memory, the rest is just supporting electronics.

In one of the cabinets of the MIOM row there was the main power supply. This was also built to last the ages. Big copper wires, big switches:

For the rest, there was just many cabinets filled with wiring. Not simple wiring, many of these wiring connections were made by hand and the dread of any technician. If anything went wrong with these it was a matter of getting a Logic Analyzer in there and simply check the system to see what went wrong. Once that was found it was a matter of tracing every single wire to see if any of the connections were bad.

The system had 7 tape drives, two of these were in constant use and needed to be able to start the system. These were the statistics tape which was used to keep trackof which company sent how many to messages to which destination. The other was the history tape needed to be able to keep the 'store-and-forward' system running.

When the system went down a loud alarm would go off and the operator would first check which unit had stopped. This could quickly be done by looking at the alarm box which was locate close to the tape drives

For the actual startup a third tape was always needed, the so-called PLT, HLCPLT or Program Load Tape.

Tapes would come in all sorts of sizes, small ones for the program tapes, large ones for statistics and history tapes:

The DS714 had 7 large free standing cabinets which each held one tape drive, the DS790 was more advanced and had drives which fitted with 2 in one cabinet.

There were a few self loading tape drives. These were (in theory) a blessing to the operator because you could simply slide a tape in and the drive would do the rest of the work.

Sometimes they would give some problems though:

Specially the self loading tape drives had to know where the start of a tape was. This was indicated with a BOT (begin of tape) which was a little piece of reflecting material which was put a certain distance from the physical end of the tape

On top of the tapedrives you could find a wooden block which black labeling. The operator in the operations room could see the tape drives and with the aid of these blocks he could in a glance see what each type drive was doing:

To clean the drive's read and write heads, bottles of pure alcohol could always be found in the area of the tape drives. This stuff was very much not for drinking but it was used for all sorts of cleaning: