A PDP8 Computer Room probably Georgetown

Joe Brandon who worked for Autotote, Automatic Totalisators' American subsidiary company, made the following comment in 2015 on seeing this image in the photo gallery. I see you've added tons of stuff. Not only as to the J8's, but I was pleased to see the picture of a computer room, with the negative logic scanners. Talk about an acid flashback for me.

I worked on the totalisator systems that followed this generation. This tote is based on three PDP8 Mini Computers. I worked on the PDP11 Mini Computer based totes and later on the VAX11 Mini Computer totes, whilst with Automatic Totalisators. Although the machinery in this room is quite familiar to me, despite not having worked on these systems, something that I am closely attached to in this photograph is in the top left hand corner. There is a CRO (Cathode Ray Oscilloscope) standing on a trolley. It looks like a Tektronix valve based CRO. On the left of this is a smaller CRO, which looks like a Tektronix 465 which is a transistor based CRO and consequently more compact. It is facing upwards angled forward to reveal the front of the CRO. Much of my time with Automatic Totalisators was spent using these to analyse circuit fault conditions to identify a faulty component so it could be replaced restoring the circuit to a functional condition.

This image highlights a feature of all ATL electronic totes. The world's first computer based tote, which ATL developed for the New York Racing Association, as mentioned above, is regarded as one of the first Non-Stop-Systems. For the layman, that means disruption to service is intolerable. To satisfy this requirement, the systems have to be fault tolerant. In other words they have to continue to function normally, despite any equipment failures. The way this is achieved is to eliminate all single points of failure by providing redundant equipment. That is why every ATL tote system was a duplex or triplex system. In other words the central tote computers consisting of the transaction processors and front end systems were duplicated or triplicated.

The image above is an example of a Triplex system. At the core of the tote system are the transaction processors. There are three in this system, the PDP8s and their consoles can be seen amongst the right hand equipment racks above the blue work shelf below the brown access doors. All other access doors are blue. All the computer totes I worked on were Duplex systems. They were designed in a master-slave configuration operating in a mode called hot-standby. In the Atlas range, based on VAX11s the master slave relationship was configurable. For instance, bets could be authorized for issue, when the transaction was main memory resident and written to the disk based transaction file on the Master only, or after the transaction was secure on both the Master's and the Slave's disk subsystems.

My understanding of the Triplex system, is that the third computer remained in a mode called Catchup, updating its databases by chasing the on-line computers. If the Slave fails the Master continues operation as normal and the third computer catches up with the new Master to become the new Slave. If the Master computer fails the Slave ensures all main memory databases are up to date on its disk subsystem and responds to retransmissions from TIMs resulting from the Master not having completed transactions initiated before the Master failed and then the Slave becomes the new Master. The third computer then catches up with the new Master to become the new Slave, exiting Catchup mode. If the failed master can be repaired it can be restarted in Catchup mode to take over the role of third computer.

The row of J11 TIMs in this photo are presumably in the computer room for testing or alignment purposes. They will of course be deployed to selling positions around the track when in use.