The Burning Ends: Collecting Sunsets

Speaking of Funerals...

Oh boy, here's an interesting if macabre topic. It's actually quite appropriate coming after my last post about letting go.

As I've gotten older, I've started seeing end-of-life events. Not talking about humans here, we all have to deal with that, but about systems and businesses. It's something that as a young IT person you don't think you'll ever see, but then Father Time comes by and whacks you in the nuts with a baseball bat.

I don't think I've ever been involved in an end-of-life for any systems I helped start, or where I came in for some work during the middle. But I have now seen a few in just the last couple years where I've gotten involved in a system during its twilight years.

There are a couple of distinctions to make about these first.

One type of end-of-life is witnessing the actual death of an application or system. It's dying and going away for good.
Another is seeing systems die but only in a specific implementation. The software may still be alive elsewhere but is standing down in an installation, perhaps as a part of a company closing.

Total Death

The first end-of-life I saw was a true death of a system I supported for an oil and gas company. It was a home-grown system built around 1997. It handled some business for this company until 2022. Twenty-five years is a respectable life for a corporate system, often some of the world's worst written and most poorly designed systems. This app had an interesting life; the software was heavily used and had a team of 15 people at one time supporting and enhancing it. About 2012 the company split and the software was still being used for about half its functions. I was part of a team brought in to keep it and a sister system (a third party software package) alive during the transition. We were probably expected to just be there a couple years and before the client migrated to a new system.

This system was proprietary, old, clumsy, and ugly. No one wanted to work on this bitch. The full-timers at the company wanted nothing to do with it, and everyone on our team preferred to work with the sister system since it was an industry standard application that was to continue use long after this system was replaced. Well, guess who got assigned to take care of it? Yours truly. But I'm the kind of consultant that takes care of the customer, even if it's a career limiting move, because that's what I'm being paid for. I hope, probably in vain, that the qualities of loyalty and reliability will be as useful in garnering future opportunities as simply having the latest acronyms on the resume. So I took it on and became a one-man show, handling all the support and even improving the software some. From a team of fifteen to a party of one...not sure if the client appreciated that but I absolutely provided value.

Real life threw a curve ball at this company and a downturn in the energy industry around 2014 caused the company to turtle and lay people off left and right. Everyone on our team got terminated except for the one person that knew how to keep alive the system that was still managing business and producing invoices. Again, yours truly. And it wasn't just contractors being let go, I outlasted the dozens of full-timers that didn't want any part of this system. This taught me a useful lesson that's not a tautology, but can be a useful tactical move: sometimes it's good to be good at the job no one wants.

Things picked up by 2016 and some of my team was brought back. But between budget cuts and project delays, I last another six years, for a total of eight, with this client. The sister system was upgraded and picked up most of the business the proprietary system handled. The proprietary system was finally feeling death creep closer. I'm actually released from service in late 2020, so I miss the actual system shutdown that happens in 2022. But I was close enough to capture the essence of it; activity slows to a crawl, there are no enhancements to be worked on anymore, and even support dwindles. Yes, frankly it got a little boring at the very end, though it was certainly less stressful than when things were busy.

The client mothballs the system for archival purposes and ceases using it for daily work. The system is then buried under layers of time. Many people that worked with this system would say "good riddance," as do I, but also I feel a little sadness. It's like all the creativity and work of a generation just becomes a tiny footnote in the company's history...and maybe not even that. Just gone. If systems had souls, I would say to this one, you helped some people in your life. That's meaningful even if it's forgotten.

Partial Death

The second story is of a system's sunset in a specific implementation. On one contract, I'm brought in to join a team that's supporting the systems for a company that's just been sold to a holding company and everyone knows it's going to be sold again. We don't know when so we carry on as professionals and do a good job knowing it could end at any time. It does, about two years after I sign on. The first year I was there was pretty busy; we had ongoing business and multiple systems to manage, and we also did some upgrades and generally carried on like a regular business. In the middle of the second year, the sale of the company is announced. Any enhancement or upgrade projects are instantly canceled and we shift into support-only and business transition modes.

In this case, the products don't die; the two key systems we supported are both third party products that as far as I know are thriving in the market. At this company though, they're going to cease functioning. Not so sad for the applications, but more for the people side of things as the group is going to be broken up. The staff splits between a group that transitions to the new owner, a group that voluntarily leaves, and the ones that would have liked to stay but are asked to leave.

No matter how it happens, it is as DeMarco and Lister said (and I paraphrase slightly in brackets): "Whenever a system goes live [or dies], someone gains power and someone loses power."

Closure

Why did I choose to write about this today? I think it's because it's September 29, 2023, the last day of Netflix's DVD operations, known as DVD.com. The service pioneered internet-based service companies twenty-five years ago and was also a leader in the transition to streaming and taking a hand in content production. The DVD side of the business naturally dwindled as streaming's convenience looked like a tall and handsome alternative to a short, ugly, and older option, and humans are pretty predictable in those circumstances. A few of us holdouts still appreciated the qualities that the physical media service offered, but even we had to know it wasn't going to last forever.

Netflix of course, is still a Wall Street darling and is doing well, but I can't help but feel a little nostalgic as I remember the DVD service's halcyon days and how much fun it was to add movies to a queue and have them appear in the mailbox a few days later, and at a much more reasonable price than what the buster on the block was charging.

In my opinion, the drawdown of the Netflix DVD business is a stellar model of how to sunset something.

First, communications went out months in advance to notify customers this was coming.
Then, the website was updated with a FAQ pre-emptively addressing concerns customers would have.
The service continued at a high level of quality and professionalism until the end.
On the last day, DVD.com even sent out parting gifts. The company sent a few extra disks from the titles still in the customers' queues, and clearly explained that any disks customers had in their possession past September 29 could be kept if desired.
Finally, the company executed a graceful departure, issuing thanks to customers for their patronage.

This is an end that was sad, but also well executed and one that didn't leave behind unsatisfactory mysteries or a sense of being ripped off, as can happen when a magazine shuts down and subscribers are left with unfulfilled issues and no refund. This was an end with a comfortable closure, so much better than the sloppiness of America's departures from Vietnam or Afghanistan.

Here's a capture from the site as it was on the last day.