Well it was a guess. That's why I qualified myself with "I may be full of it". There isn't enough information about their system to make a serious assessment.
Maybe they did. Maybe they decided it's possible but would take too long and they can't afford the loss meanwhile. Maybe they explored options but don't have the right knowledge.
But 10k€ monthly cost for serving 6 million users seems avoidably high unless it's something compute and data intensive like gaming.
It seems like a perfectly normal cost, even a sensible choice, for a cloud-based, invested-in startup with cash to burn that is optimising for speed to market and growth. But not for one that is cost optimised.
All that said, I wonder if their cost is actually mostly on people to run and develop the thing. Other comments have taken it as meaning the cost of infrastructure, and I ran with that. But soup.io's own note does not say it's all on servers.
If that's the case, obviously it's a different situation and there may be no reasonable way to reduce the costs below revenue.
Is it possible to design a piece of software that is identical to soup.io but runs at a fraction of the cost?
The answer is almost certainly "yes, if you're willing to rewrite it from scratch". Unfortunately the capex required to do that weighs heavily on the opex saving.
"They should have designed it from the start to be efficient" would normally be my next thought - but there by grace go us all.
I'm usually pretty quick to dismiss such claims as well but in this instance I agree with GP jlokier. I do these kinds of massive emergency cost reductions fairly often with my clients and there's always a story of the current maintainers either missing some critical knowledge about what they're currently using / could use instead; or simply massive tunnelvision.
Some key factors here:
- Discontinuing expensive parts of the product is better than discontinuing the entire product.
- I see the M word being thrown around and that… uh… potentially says something.
- "We're not open sourcing because the product is too complicated" is also extremely telling.
- VERY often, "I'll discontinue because I can't afford to run it anymore" hides an underlying "I don't want to run it anymore"; one the maintainer sometimes doesn't fully realize themselves. I've seen this a lot on GDPR day, people shutting off services because it's "too expensive to comply". Talked to a bunch of them, and after a lot of chatting it always boils down to "This will give me a much needed break from the stress of running this thing which doesn't pay my rent, and I get to dodge the blame".
I'm going to go ahead and extend the offer GP can't make. soup.io maintainers, if you're reading this, are indeed spending 10k+ EUR on infra, and do want to keep your service alive and running, please reach out, I'll work pro bono. I also have some good contacts in the archiving world if it comes to that.
>I'm going to go ahead and extend the offer GP can't make. soup.io maintainers, if you're reading this, are indeed spending 10k+ EUR on infra, and do want to keep your service alive and running, please reach out, I'll work pro bono.
If you do this, and you manage to make significant process, I'd love to read a writeup of it.
If you're doing the work pro bono, then the least they could do would be to allow you to generate some content for yourself to use for self-promo.
Your last point is vital. I've negotiated to cost cut services for people several times where the sticking point in the end was that they just wanted out. One of them ended up giving away a site I could have made highly profitable in days in order to be able to cut and run.
From my experience evaluating projects like this: Maybe, but many don't. There's decidedly a market in identifying services like this and taking a cut from making them profitable. If I ever go back to contracting, it will be the market I focus on - so many low hanging fruits.
> Do you think they might have explored options to reduce costs before deciding to discontinue the service?
Possible. But they might not be very good at it if they’re serving 6M users with 10K/month in cost in the first place, so they wouldn’t know where to get the wins either.
Of course, it’s also possible us armchair devs are just wrong.
