When is Technology ‘Legacy’?

I recently got into a discussion with someone about what made a technology ‘legacy’. This happened because I pointed out that the typical enterprise Java / .NET infrastructure was effectively ‘legacy’ technology with the rise and rapid adoption of new serverless architecture patterns, containers, etc. The person I was speaking with disagreed; saying that mainframes were ‘legacy’, but the big stack Java app servers and such were still ‘modern’. The ‘new’ patterns were just that ‘new’ or ’emerging’.

That got me thinking a bit about our terminology. The Java / .NET infrastructures that sprang up over the last 20 or so years were the ones that proved the mainframe was ‘legacy’ as they became the ‘go to’ distributed approach after the Client/Server and distributed computing revolution of the early to mid ’90s. Java, as a language, is now well over 20 years old. That is [roughly] the age of the COBOL / mainframe when it began to be seriously disrupted by the dawn of new patterns running on mini- and micro- computers. There was broad agreement that the COBOL / mainframe pattern was ‘legacy’.

Now, we are disrupting the ‘standard’ enterprise pattern all over again, but the term ‘legacy’ does not seem to be moving to keep pace. It reminds me of the odd duality of what an ‘antique’ car is versus a ‘classic’ car. The Ford Model-T has been called an ‘antique’ for a very long time and would have been referred to as such in the mid-1960s. At a time when the newest Model-T was not even 40 years old (the last Model-Ts were built in 1927). So, in 1965, they would have been only 38 years old. By contrast, the Ford Mustang from 1965 is a ‘classic’ in common parlance today – despite the fact that it is 53 years old.

There is definitely something very human about it. It probably has something to do with sentimentality to a given technology or technological era with which people connect. I will leave that to others to debate. I am empathetic with such sentimentality – I do love the design of cars from the mid-50s through the 60s – but, I would not drive one every day as my primary transportation. I look at computers through a similar lens. I respect the patterns that got us here and have a high level of historical respect for them, but I have little patience for obsolescent technology or those who cling to it. Which is what led to my comment about the typical ‘distributed’ stacks in most enterprises. In my view, dealing with what you have while you figure out how to replace it is very different from clinging to something that needs to go away.

In the end, of course, it is subjective and down to perceptions born of organizational culture. Does it help your organization move things forward to call the existing infrastructure ‘legacy’ regardless of its age? Is it meaningful or just a label used politically to try to get the organization to invest / change to new technology? We are all certainly going to need to answer these types of questions for ourselves and our organizations as the technology paradigm shifts.