Comment by voltagex_
8 years ago
I saw someone post on Twitter saying they were tired of being their own sysadmin.
It's true - I could set up my own mail server, my own Git hosting etc etc, but why would I? I'd rather just pay Fastmail for email and GitHub for some private repos.
I'm my own sysadmin (for personal email, small HTTP server, an OpenVPN and a few other minor services for personal use) and while it did take me a non-negligible amount of time to get things running smoothly it now only requires a few minutes of attention every other week to apply security updates and make sure my backups are running smoothly. In exchange for that I have total control over my data and basically unlimited customization potential. I'm definitely not planning on stopping that.
I do use Github regularly however, but it's because of the "social network" aspect. If you want to interact with many open source projects it has to be on github nowadays. The network effect is strong.
Congratulations, you can configure your own mail server. You're already one in a million. Out of that group, maybe one in a hundred can do it nearly as securely as one of the big services. This is the reason why people use the big centralised services, and also the reason smaller ones don't survive: email is ridiculously over-complicated. You won't convince anyone it isn't by explaining that you managed to DIY in under a week.
Fortunately, rolling your own git isn't nearly as hard, and the percentage of people who could run a node compared to total users is larger. Hell, every person using git also has a fully-functional, secure and standard self-hosting stack already installed. Getting a GitHub-like web interface is similarly easy. This fundamentally reduces the risk of centralising, which conversely means that people don't have any qualms about using GitHub, since they can leave at any time. (Less true of issues, PRs, community, but hey.)
In either case, centralising has large benefits: one for handling complexity, one for giving you a community.
When you want decentralisation, it's because you don't want a single entity controlling what the ecosystem becomes. The circumstances where this is true are typically big utility-like things, where we care more about guaranteeing that water and electricity exist than the top end of providers' profits. Email is like this. You would never build your own power station, just like these days it would be bordering on negligent to use anything except Office365/GSuite/etc as your corporate email host. Lots of jurisdictions will create highly regulated marketplaces so you don't get Enron-California-style anti-competition. When things get that bad in Gmail-land, let us know.
[Edit: I was pretty snarky just there, but not at you! It was in general, I swear.]
I want to make that switch, but am rather intimidated by my inexperience in that domain. Are there any guides or tutorials you've used that you'd recommend?
Not a single tutorial comes to mind but you generally find a wealth of resources and tutorials online about setting up anything you might need.
Always start by setting up a strict firewall and then whitelist cautiously anything you need as you set up the rest. Don't bother with SELinux. Be very careful with your Postfix (or if you're masochistic, sendmail) config as it's rather (needlessly IMHO) intricate and it's easy to end up with a configuration that appears to work but is extremely broken. In particular if you end up configuring an open relay by mistake spammers are going to have a lot of fun with your server but you'll end up blacklisted everywhere.
My "stack" is nginx for web, postfix + dovecot + spamassassin for email (+ roundcube when I wanted a webmail). Postfix was by far the trickiest to configure, although if you find a good tutorial online and can spare a couple of hours to understand how it all works it's not too bad.
1 reply →
Have you tried mailcow? I find it strikes a pretty good balance between running your own mailserver while not getting overwhelmed with details.
It runs in containers and provides a webserver, webmail, caldav, spam filter, updater... I have made an ansible role that will back up to a Borg repository every hour as well as restoring the latest backup upon installation if you need inspiration: https://galaxy.ansible.com/coaxial/mailcow/
If you use your own domain name, you can set mailcow to sync with your Gmail account so you have all your emails in mailcow and then pull the plug on Gmail without losing anything. I find it runs pretty well for me on a server with only 2gb of ram.
If you are on Windows, the equivalent is smartermail. For a fee it also supports ActiveSync, which simplifies client configuration and gives pseudo-push.
Decentralized =! inconvenience
Except that, 99% of the time, it does.
For the same reason that you don't opt for a dictator even though it is more convenient than democracy (until it isn't, at which point there is little you can do about it).
Also, you could also instead pay people to develop easier solutions for self-hosting. Or you could buy a service from a smaller hosting provider. There is more than the extremes of "google is email" and "I have to hand-carve my mail server".
So you are suggesting it is a moral obligation to choose a smaller provider, even if the largest provider is better?
Centralization happens because someone ends up doing it better than everyone else, and so everyone chooses to use that provider and they become the dominant force.
I get the need for diversity, but as an individual, I am going to choose the best provider, even if they are the biggest one.
> So you are suggesting it is a moral obligation to choose a smaller provider, even if the largest provider is better?
Not moral but, I’ve noticed that things all seems to be tied to a pendulum that swings back a forth. Pushing in the opposite direction is required to attempt a balance.
Why is the larger one better? i use email mainly over IMAP and can't see the better in using Gmail. Yes, you should prefer smaller email provider. We live in the cyberpunk world now, corporations are the ones with power.
> So you are suggesting it is a moral obligation to choose a smaller provider, even if the largest provider is better?
Maybe, but the world really is too complex to break it down to such a simple question, it obviously is a tradeoff with many more factors to consider.
> Centralization happens because someone ends up doing it better than everyone else, and so everyone chooses to use that provider and they become the dominant force.
Well, but does it? I mean, no doubt such cases do exist, sure, but if you really look into how companies do become dominant, that is only one of many factors, and sometimes not even a necessary one.
> I get the need for diversity, but as an individual, I am going to choose the best provider, even if they are the biggest one.
Well, but how do you evaluate what "the best provider" is?
You might be comparing functionality, say. Or price. Or speed. Or any other property of the product as you could now choose to use it. And obviously all of those are important things to consider.
But my suggestion isn't that you should follow some abstract moral teaching because some ideology says that this is the right way, and the only right way. My point is that it may even be in our very own interest to choose a solution that is inferior in terms of current functionality/price/speed/whatever because there are long-term costs attached to the superior solution that actually make it more expensive, all things considered, than using the inferior solution now. So, arguably, the currently technically inferior option with a lower total cost would actually the better solution.
To maybe make it more practical, but without any claim to being realistic, the numbers are obviously just made up: Let's assume that using Gmail saves you 10 minutes every day vs. using Thunderbird. Now, Gmail is privately owned, so if everyone chose to use Gmail, they would effectively have the monopoly over email. At that point, they have every incentive to add proprietary functionality for the sole purpose of making interoperability difficult. Which could prevent a new competitor from entering the market what would invent a new email workflow that would save you a further 10 minutes every day. Now, does choosing Gmail actually save you time overall? And is Gmail the better product if using Thunderbird now would lead to you being able to save 20 minutes a days a few years down the road?
This isn't about some sort of diversity for diversity's sake, this is about which of those options actually is in our very own long-term interest, and monopolies have a strong tendency to be very much not in the interest of the customer.
So, really, if anything, I would suggest that there is a moral obligation to watch out for people/organizations accumulating too much power and to prevent them from obtaining it if the long-term damage that that concentration of power can do is worse than the short-term benefits obtained from using their offerings.
2 replies →