Datapac's Demise

I'll start the new year with a look back at the national data network that predated the internet: Datapac. The most surprising thing to me about the recent demise of Datapac was finding out that it was still operational into 2009. I knew that for the longest time it continued to be used for point of sale (POS) and bank machine (ATM) transactions, but I hadn't thought about it for years. It seems they kept using it by paying attention to the maxim that tells us: if it ain't broke, don't fix it. Like the SS7 telephony signaling network, Datapac was very difficult to hack into since there were no or few connections to outside networks from which attackers could attempt entry.

Datapac is (or was) an X.25 network. X.25 was the closest we came to publicly-accessible data networks before the internet was opened to the public. It was never cheap so its use was pretty much strictly commercial. Like IP, X.25 is a packet-oriented protocol. There were many X.25-based network around the world, and they were interconnected through gateways that had to (usually) be manually navigated to reach a desired destination: the X.25 address space was small and the various networks did not and could not coordinate assignments.

Datapac has some special significance to me since I worked with it back in the late 1970s when it was shiny and new. I wrote a software package that interconnected my employer's one Datapac connection to any of multiple stations in a time-sharing minicomputer installation, making this valuable resource widely-accessible to authorized users within the company. For a young, fairly-inexperienced programmer this was a tremendous learning opportunity. That was my first exposure to highly-concurrent software design, including interrupt drivers and low-level resource contention, in a live commercial environment. It also taught me communications protocols, which came in very useful later in my career.

X.25 itself became obsolete years ago as the technology improved. An important attribute of X.25 is its ability to get the data through when the communications links are noisy and unreliable. With the wide deployment of digital carrier systems and increasingly-sophisticated modems, the overhead of X.25's data link protocol became superfluous. Doing away with that overhead was one of the design goals of Frame Relay.

Of course today pretty much all communications protocols other than IP are in decline. Nevertheless, it is important to occasionally look back on what came before and reflect on what those earlier generations of technology gave us. In its time, Datapac served its purpose, and it served it well. The torch has been passed.