3 ms·
Amusingly, there was a commercial alternative in the same period - Tymnet. Tymshare, a computer time-sharing company ("the cloud", version -1) had machines at v
by Animats 7y ago
Amusingly, there was a commercial alternative in the same period - Tymnet. Tymshare, a computer time-sharing company ("the cloud", version -1) had machines at various locations and wanted to connect them. So they had their own packet switching network, mostly for terminal to mainframe connections. Control and route setup were centralized (this is now called "software defined networking") but packet forwarding was distributed.
Tymnet was very cost-effective. Its purpose was to share expensive 9600 baud leased lines (hundreds of dollars a month) among multiple Teletype machines running at 110 to 300 baud. Worked fine. Remote echo was rather sluggish, as echoed characters tended to come back in groups of four characters. That's because they used an accumulation timer approach - wait until you have some minimum amount of data or the timer has run out. This is terrible for latency. (The current version of that is the "delayed ACK" timer in TCP, which I consider a mistake.)
There was a real question in the early days over whether networking would be packet-switched like the ARPANET or circuit switched like Tymnet, X.25, and SNA. I thought at the time (1970s) that circuit switching would win out, because packet switching cannot deal well with congestion in the middle of the network. I didn't expect fiber backbone bandwidth to get so cheap that the congestion stayed at the edges.