“Avoid vendor lock-in” is the dominant architectural advice. It’s wrong about half the time. There are situations where picking a stack and committing is the right call.
When lock-in is genuinely a risk
The vendor has clear pricing-power leverage. Your migration cost compounds with time. The vendor’s commercial behavior is hostile. Concentration risk is regulator-relevant.
When lock-in is overrated
Your engineering team is small + your platform sophistication is moderate. The cost of multi-vendor abstractions exceeds the migration cost you might face later. The vendor’s product roadmap aligns with your needs and they have skin in the game.
The decision framework
Three questions: Will I want to leave in 5 years? What does that cost in money + time + risk? What does multi-vendor abstraction cost me today + over the next 5 years? If today’s + accumulating cost of abstraction exceeds the future migration cost, pick a stack.
The “always avoid lock-in” advice is a meme. The real answer is contextual.