Legacy Systems: The VHS Timescale
The last standing manufacturer of VHS video recorders has announced that production was to be discontinued. “Finally” I hear you sigh!
The technology has been around since 1976, and could have been considered obsolete as early as 1996, when the DVD was first introduced. Even though a more convenient, more permanent, more versatile, better quality alternative was available, VHS videotapes have lingered on for another 20 years just because they were there, just because of this awful thing called momentum.
The end of the production of recorders does not even mean the end of VHS as an ecosystem. There will be a second hand market. Aficionados will scrutinize eBay for players in mint or at least functional condition, or even defunct ones for spare parts. There will be VHS owner conventions, with funny T-shirts and bumper stickers to share the love for the mighty cassette with the world.
In a nutshell, the VHS technology has been up to date for 20 years, obsolete for 20 more and will probably survive for at least another decade.
And VHS is not alone out there. Technologies outlive their relevance by decades. The French Minitel, which turned France into a high-technology telecom power, was introduced when Jimmy Carter was President of the United States, and discontinued in a time crowded with IPhones and Facebook.
But VHS and Minitel are easy cases. They can both be replaced without too much of a hassle, with incremental deployment, interim solutions, or any other slicing or risk mitigation strategy.
In stark contrast, replacing a non-trivial software system, a programming language, a database or transaction processing engine, a hardware platform or an operating system is significantly more complex, and this complexity further extends the timeline for their demise after they have become obsolete.
Still, people often wonder at the number of these archaic technologies still in widespread use in the professional IT world. When I was peeling my first bananas in this monkey business, common wisdom was that COBOL was about to disappear. As a graying gorilla, I now know for a fact that it will outlive me.
So will the mainframe.
They are both substandard compared to available alternatives. They are both expensive to operate, and the skill shortage is making them more expensive over time. Despite the obvious incentives, replacing them is so challenging that all the efforts at doing so have merely skimmed the surface. Thirty years since the first mainframe replacement products arose, the bulk of the installed base is still to be migrated.
The choices we make have lasting consequences, and the timescales are frightening long. Ironically enough, the more successful the system is, the more likely it will be to stay for long, and the more painful will technical dependencies becomes.
COBOL and mainframes are still part of the common IT vocabulary, but what about PL/I, Mumps, Model 204, RPG, Focus, and dozens more languages and databases you probably never have heard of? They have been in used for decades at organization that depend on them, so dramatically that they can’t afford to survive without them. They may still be supported so that the installed base is not left in the dark, but their very paradigm is no longer relevant.
When these technologies were selected, no one thought these systems would still be used 40 years down the road. Long term availability probably was not even a decisive factor, compared with personal preference, technical adequacy or price.
And here we are now, with big organizations depending on obsolete technologies that must be kept alive at any cost. The fast-paced world of startups has changed many things about how to do business, but not this timeline expansion issue.
Most startups are oblivious to whether their underlying technological choices are sound in the long run or not, because they don’t live long enough to care. The exceptional successes, the unicorns turn into more common corporations, with similar concerns.
Consider Facebook, for instance. Before it turned into the behemoth we all know, it began with a startup state of mind, driven exclusively by time to market and fast delivery cycles. Processing power was cheap, labor expensive, and the tools of the trade were about flexibility and convenience. PHP was a reasonable implementation language.
Fast forward to a billion users, data centers all over the planet and an influence that dwarfs governments, this entire ecosystem still runs on PHP, but when cycles are run on tens of thousands of machines, CPU power is not that cheap any longer. These performance issues were not serious enough to have Facebook reconsider PHP as a strategic choice, they went for a radical solution instead: they started a large scale research project named HipHop for PHP, to develop a compiler and virtual machine that would use static and dynamic analysis to alleviate the performance penalty induced by the polymorphic and interpreted nature of the language.
This was motivated merely by these performance issues, but the HipHop project has bought Facebook something at least just as valuable, namely independence: they no longer depend on any external organization for their core technology. Whatever PHP becomes, wherever it goes, even if it disappears altogether, Facebook can keep functioning.
But what about normal-sized companies with normal-sized pockets, that cannot afford maintaining their own core technologies? How can they survive? How can they live with decisions made in another millennium? What about those that depend on something larger and more complex than just a language? The story about VHS recorder owners scouting eBay for spare parts is not just a colorful image: we’ve seen companies do exactly that for hardware components that were no longer supported by their original vendor, and which would have put their business to a standstill if they were to break down. They bought machines with less processing power than a modern washing machine for the price of a brand new family car, leather seats, audio system and all. Every machine in working order they could find felt like a victory over time.
You can entertain the delusion that these people were insane, that it just could never happen to you.
You are making choices today, based on what you know of the world today. How the world will be in twenty years, how your choices of today will impact your ability to function by then is a different matter altogether.
P.S.: Interestingly enough, the size of the market does not say much about its resilience. VHS displaced, crushed, and obliterated the incumbent Betamax, pushing it into irrelevance (down to 7.5% of the US market by 1986). Still, Betamax-compliant recorders have been produced until 2002 and the production and distribution of tapes got discontinued in 2016. Go figure.
P.P.S. And by the way, don’t get me started with our tendency to accumulate videotapes for the sake of owning them. How often do we actually enjoy the faded colors and the cracking soundtrack of these dusty relics anyway? The same goes for photographs, only made worse by the advent of their digital variant and smartphones. People now keep gazillions pictures, thousand time more than they‘ll ever be able to watch, as if life was more about the narrative in images than living it for real. And the stock is still growing. As if the world needed another selfie, or another depiction of the Taj Mahal.)