A few years ago I helped a team adopt a proprietary cloud database that made everything wonderfully simple. It scaled itself, backed itself up, and turned a week of infrastructure work into an afternoon. We were delighted. Nobody in the room asked the one question that mattered later: if we ever want to leave this, what does that cost us? Two years on, when the pricing changed in a direction we did not like, we discovered the answer. Our data model, our queries, and half our application logic had quietly married the vendor's particular way of doing things, and the divorce would take months.
That is the trade at the heart of every managed service, and it is a genuinely good trade right up until it isn't. You are renting away a problem you do not want to own. In exchange, you accept a dependency you cannot easily reverse. The convenience is real and immediate. The cost of the lock-in is invisible and deferred, which is exactly why teams keep underpricing it. Nobody feels the exit cost on the day they sign up. They feel it years later, at the worst possible moment, when leaving suddenly becomes the thing they need to do.
I am not against managed services. I use them constantly, because reinventing a database or a queue is rarely a good use of anyone's life. What I have learned is to go in with my eyes open about which parts of my system I am tying to a vendor and how deep those knots go. Using a managed queue through a standard protocol is a shallow commitment. Building your core data model around one vendor's proprietary features is a deep one. Both can be right. The mistake is not knowing which you are choosing.
So before I adopt something I cannot easily walk away from, I make the exit cost explicit while it is still cheap to think about. What would it take to move off this in a hurry? Is my data in a format I can export and actually use elsewhere? Am I depending on a standard interface, or on this vendor's unique flavor of it? None of this stops me from choosing convenience. It just stops convenience from choosing for me.
Lock-in is not a reason to avoid a service. It is a price to name out loud before you agree to it, because the easiest door to walk through is often the most expensive one to walk back out of.
– Serguey Shinder
Top comments (0)