[Eridrus]

I think I have a personal sense of unease when pieces of my systems are proprietary, very hard to replace, and under control of a different entity. But it's really not my primary concern. My primary concern is moving quickly and not spending too much money.

I am a satisfied AWS MemoryDB customer. It's not open source, but I feel more comfortable that it implements a semi-standard protocol (Redis) so if I needed to get off it I could.

I am also a satisfied customer of Doppler. I briefly looked at Hashicorp's Vault product, but that was just so complicated to use I think the choice of Doppler (or one of their competitors) over Vault is clearly correct for startups. This would be a pain to replace quickly, but fundamentally not that big of a deal, it's just a reliable key value store with a bunch of integrations.

I am still a cheapskate though, like every other developer, we got a quote for Dagster+ for 20k/yr recently, and decided we'd rather keep using self-hosted Prefect.

I think the main takeaway I have from selling software is that selling it to front line developers is a fool's game unless it's cheap and you're going for huge scale. If you want to charge more, you need to sell to someone who has cash, but very indirect leverage over development.

[aambertin]

That's gold, thanks! So, what I understood from the MemoryDB story is... value-for-money ... and try to avoid lock-in. But -from the Doppler bit-, a "soft" lock-in is acceptable if the price is right.

Did I grossly misread something?

[Eridrus]

I guess I don't feel super locked in to doppler either; everything takes creds via env vars these days, so it's not even that locked in. But there would be a bit of work to migrate thing, etc, like with everything.

Good value is probably the most important thing; the value could be speed or efficiency or whatever, but that's obviously the first bar.

Open source is not high up on the list of questions.