Marco.org

MG Siegler laments Gmail’s recent issues:

Google, please set a price — any price — that you determine is necessary to keep anyone’s account running smoothly at all times. I’ll gladly pay it. I don’t care if it’s $100 a year or $1,000 a year. It would be worth it.

People often talk about the desire to pay for Twitter either for better uptime or for more features, but the situation with Gmail is much more serious. Unlike Twitter, I conduct basically all my business through Gmail. I simply need it to work for me at all times. And I’m happy to pay for that to be the case.

I’ve seen many similar pleas recently whenever any popular, free web service has problems: “Please, let us pay you so there won’t be any problems!”

But it’s an impossible dream. If a web service is popular enough that you hear about it when it has downtime or major issues, it’s probably a large, very complex system. 100% uptime is effectively impossible. There are far too many moving parts that fail, resources that run out, boundaries that get crossed, and bugs that make themselves known at inconvenient times.

The operators of such services jump through hoops behind the scenes to make sure that bugs don’t get shipped, failures are routed around, and expansions happen behind the scenes without bothering the users. But nobody’s perfect, nobody has infinite resources, and nobody can predict every problem before it happens.

That said, there’s never any guarantee that a service that has been good in the past will always be good in the future. Siegler’s (and TechCrunch’s) problem isn’t that Gmail has been unreliable (which really isn’t new), but that there’s no good alternative once you’ve invested heavily in it — either by giving out a @gmail.com email address, depending on features that other providers don’t support, or growing accustomed to (and dependent on) the Gmail web interface.

For something as important as email, I’ve never trusted everything to a proprietary provider. My email address has never ended in someone else’s domain name, and has never been hosted in any way that would preclude me from easily switching to another provider.

Since 2007, I’ve used FastMail, a paid IMAP host, for my @marco.org email addresses (with the $40/year “Enhanced” personal plan). Rather than using its (unsophisticated) web interface, I use Apple’s Mail app.

FastMail’s uptime has been incredibly good — I don’t remember the last time I saw any downtime, but I’m sure that the total downtime I’ve ever seen from them has been less than an hour. But if it ever starts to suck, it’s just IMAP, and I own the domain, so I can switch to any other IMAP host easily (or self-host it, which I don’t recommend, but it’s always an option).¹

All of my messages are downloaded by Mail and stored as files locally, so if a data-loss disaster were to happen at FastMail (which can happen to any service, even Google’s), I can recover my email from my personal backups.

You must own any data that’s irreplaceable to you.

By relying on a hosted service with no direct alternatives or difficult outbound migrations, you’re giving up a level of control that you shouldn’t for something as important as your business email.