When Tymnet decided to provide remote terminal access to computer systems, or hosts, not operated by Tymshare, which owned Tymnet, we used several kinds of connection with those computer systems.
Often we would merely install a mini computer with many cables with the sort of plug found on the telephone company’s modem.
This was easy to install for no host software modifications were necessary.
Most such host systems assumed that a particular host port (cable & plug) was for some particular speed (baud rate) and modification of that speed required human operator intervention.
Users of such hosts were accustomed to a fixed speed however.
The great advantage of this scheme was that it was very easy for a customer to try out the new service with little risk or expense on his or our parts.
This was indeed how the earliest internal Tymnet service was provided.
Service to Tymshare machines had progressed to something like DMA where the code in the base (a mini computer with communications code) was able to read and write the real memory of the host.
An early Tymnet customer was the Sumex machine at Stanford.
They wanted features that Tymnet provided to Tymshare’s hosts such as remote echoing.
These features were not available via the plug compatible interface.
They operated a PDP-10 system which was like some of Tymshare’s.
They were concerned with the security implications of our code using read write access to the real addresses in their system.
We offered the following solution to the security problem.
The hardware DMA was weakned by limiting the real addresses available to the comm code.
This was a modification to the hardware that they could visibly audit, when you could still see parts of a computer.
The comm code in the base was scarcely modified and they made modifications in their oerating system much like those we had made to our own.
We also built a channel adaptor so that a Tymnet base could appear to an IBM mainframe as a device on an I/O channel.