Not necessarily. Plenty of desktop app 'brochure' sites use a hosted solution, so zero web server maintenance required.
> And if you pick the right tech stack, those things don't tend to go down all that often.
What's "the right stack"? I'm sure Amazon and Google would love to know your secrets to maintenance free SaaS.
> You don't need "Five Nines" of availability for most types of business software, anyway. Three Nines is fine. Even Two won't generally lose you any customers.
Two nines is >7 hours a month downtime. That's basically one business day per month, the app is unavailable. Given that this is in addition to any scheduled downtime for things like upgrades, server maintenance, I'm not sure I'd be aiming for that as your baseline.
I get that the OP wants passive income from this endeavour. The way to achieve that is not to half-ass the infrastructure on a SaaS product.
I run a fairly standard Microsoft stack of ASP.NET, IIS, C#, SQL Server. Chosen for its boringness and reliability, because as I've said before, I really don't like having to worry about hardware:
http://www.expatsoftware.com/articles/happiness-is-a-boring-...
I'm also speaking from experience about the downtime thing. I used to have a rather sketchy ISP, and would see those 7 hour downtimes from time to time. I would occasionally get an email or two from customers who noticed, but never lost a single paying account as a result of downtime. Maybe it's just picking a market that can live without your thing for an hour if need be, but then that's the goal we're talking about of a SaaS business that runs itself without your intervention.
People worry about a lot of things before starting a business. They use things like "what if the server goes down" and "what if my customers call me every night at 3am" as excuses not to even try. But when you actually do try, you find that none of those things you worried about end up happening.
That's not to say it's easy. But the things you're raising as issues just aren't. In my experience.
How this is even a discussion point is beyond me.
But is that an issue? Enough to pick a distribution model for your software product that divides your customer lifetime value by ten?
If you're that worried about it, you could always take some of those profits and hire a kid to carry a pager around and do that once a year reboot for you.
My post was only a few paragraphs long, but one of them does in fact address this. I've spent maybe 400 hours this year working on my little handful of SaaS products. Zero of them were spent maintaining the server that they run on.
On the effort scale, it's nothing compared to writing the software and supporting customers.