Hacker News Reader

Comment by mattowen_uk | Hacker News Reader

Y'know what, Although I'm currently self hosting my email, my websites, my storage, my SQL, my Active Directory etc., I'm also in the process of migrating the whole lot to Azure and/or independent hosting.

Why? It's just too much hassle these days; I want my down-time to be no longer dictated by my infrastructure. I don't want to have to spend off-work hours making sure my boxes are patched, my disks are raided, my offsite-backups are scheduled, and my web/email services are running. I just want it all to work, and when it doesn't, I want to be able to complain to someone else and make it their problem to fix it.

For my data, I'll probably still have an on-site backup, but everything else can just live in the cloud, and I'll start sleeping better, due to less stress about keeping it all secure and running.

View on HN

I stopped self-hosting as soon as I moved out of university. Back in university I had a gigabit uplink and only 1 power outage in 7 years of my PhD. Now in the middle of Silicon Valley I have only 15-20 mbps and have had 3 power outages in 1 year.

Did you ever receive complaints that your emails are ending up in spam folder for Gmail/Outlook/<other big email provider>?

How about you receiving a lot of spam emails?

Nope. I'm on a static business IP, with DNS all set up correctly. I've also got SPF records set up, but I don't think they get used, as I use my ISPs smarthost for relaying mail through.

I do get a lot of incoming spam though, but I think that's more to do with some of my email addresses being over 20 years old.

Not OP, but hosting my own mail as well (postfix, dovecot, spamassassin) for six seven now. Had one issue with outgoing mails to Microsoft (hotmail I think) bouncing. The IP of my dedicated server had been blacklisted from before it was used by me, but I got them to remove it. No other issues I can think of.

I'm getting about 1-2 spam mails a month delivered to my inbox, usually french SEO spam. Not worth investigating.

Because apparently you don't need to keep stuff running in Azure patched, I guess?

Microsoft take care of patching Windows VMs, and Exchange is a service, so not your own boxes.

The author of this post cites $55/month as his cost. This is wrong. If it takes him, say, two hours a month to maintain (probably conservative) then if you value those hours at $100/hour the actual cost is $255/month.

The reality is probably in excess of $1000/month. This only makes sense for people who have an abundance of spare time, and that's pretty rare these days.

Free software for DIY hosting like this is "free as in piano." Like a huge piano sitting on the street with a sign that says "free piano," it is actually not free at all when you factor in the hidden costs.

Well that is only if you look at software/ops as a 100% commercial undertaking. It is not.

One way to understand why people self-host is to understand why people self-cook their food. It takes significantly longer to prepare food (get raw material, cut, cook) than ordering it. People still do it for $reasons - some find it fun, some find it cheaper, some find it nice to be able to control the taste, some find it more healthy to know whats going on their plate, and so on.

Only concentrating on the dollar cost is too narrow a view, IMO.

Well that is only if you look at software/ops as a 100% commercial undertaking. It is not.

Your time is only free if it is worth nothing. My time is very valuable. I happily pay other people and companies to do things for me because I'd rather have the time.

I think it's just a normal part of life. When you're young, you have more time than money. When you're old, you have more money than time.

Far fewer people cook their own food for fun versus preparation time not being the only constraint: cost, availability, health, transparency (of prep & ingredients), dependency etc

It's common for people to delude themselves into thinking they haven't wasted their time by convincing themselves they did it for fun (or the lols, or whatever) - I'd say the difference is whether they knew )or stated) this upfront, or only after they failed, or had a better solution pointed out to them.

2nd most common also: at least I learnt something / gained xp - which is fair enough, if true.

> Only concentrating on the dollar cost is too narrow a view

Not if you convey other resources/constraints in dollars. Just attach a dollar-value to your free time, perhaps with discounts for things with side-benefits.

Fair enough, by "far fewer" you basically mean the same thing I meant by "some". The point was not to have an exhaustive list of the exact reasons behind people cooking.

> It's common for people to delude themselves into thinking they haven't wasted their time by convincing themselves they did it for fun

I am probably missing some context here because this does not make sense to me. Something is fun because its fun, what does it even mean for someone to forcibly convince themselves of something that is otherwise? ¯\_(ツ)_/¯

> Just attach a dollar-value to your free time

I do that when someone asks me to do a project for them in my free time, so I can know what to charge them. But there is little value in assigning a dollar-value for time that I am going to spend doing something that _I want to do_ . Its like watching a movie, or making a sand-castle in the backyard. I won't enjoy it if I keep thinking "Damn, I just watched a movie for 3 hours, there goes $300 worth of time."

I mean the minority, which I'm not sure "some" implies. The factors you describe: cheaper, control, health can be seen as the same as cost. More expensive places might taste better, be healthier etc. While I'm sure a lot of people enjoy cooking, I'm not sure many would recreationally cook as often as they need food.

> context

You can have fun imagining the payoffs, only to find they do not appear. Have you ever seen a movie and been disappointed, or played a game and found it lacking. "fun" is not an absolute measure, and review doesn't necessarily capture how fun something is versus how fun you think it ought to be - plenty people give things higher value than their "fun" value, despite claiming to only have done it for fun - the missing value is ideological.

> there is little value in assigning a dollar-value for time that I am going to spend doing something that _I want to do_

Only if there is, some some reason, literally only one thing you want to do. But if there are competing things you might like to do then comparing them makes sense. One way or another, if you choose to watch a movie, or build a sandcastle, you are comparing the two to decide which. Using monetary values is just a more formal way of doing that for larger, less impulsive, projects.

As a developer, all of the time I spend working on hobby projects (and self-hosting has turned into a hobby) keep me up to date. It's how I learned Kubernetes, it's how I learned Traefik, nginx, and apache before that. It's how I learned how the different packaging and distribution ecosystems work for many different languages and frameworks. I intentionally host and backup some things on AWS, GCloud, and Azure. Other things live on Intel NUCs. I administer a GSuite for the family. The list goes on and on. It gives you the chance to experiment with new tools and toys that you're unlikely to use at your current job.

My long-winded point is that all of the things I've picked up have been invaluable to me at work, especially in my time as a contractor where I would be switching between many different stacks. If you want to find a "true" cost for self-hosting, you need to also treat it as training.

I don't really believe it's any different from say, a woodworker that has a shop at home. They may spend the workday just doing framing, but odds are good they find the time to make a chair, a bird house, something to keep their skills sharp.

As a developer, all of the time I spend working on hobby projects (and self-hosting has turned into a hobby) keep me up to date.

True for some things, like things that are not at all related to your work. But your job should be actively trying to make you better at your job, and a better person.

Large companies like the one I work for hire outside firms to offer classes to the employees for free, and on company time. If there is a new version of a piece of software that is significantly different from an old one, my company pays for the users to go to training, or to train online. This is very common for products like Office or the Adobe suite. But for some reason, as developers, we too often think that we're supposed to better ourselves on our own dime. If it benefits your current employer, the current employer should chip in.

I used to think this too, which is why I was self-hosting (I'm the OP of this thread), but as I've got older, and my interests have shifted, along with no longer needing to be at the bleeding edge of my skill-set (I leave that stuff to the younglings these days), I found that managing my own infrastructure felt more like a chore than a hobby, more so if it's a 'production' system and not a 'lab' environment.

"Free as in free puppy" is my other favorite metaphor. Free software is a gift to the word, but IMO it's important not to undervalue the time and expertise of operationalizing it.

It's worth remembering that you can get an expensive puppy too. I.e. choosing proprietary software doesn't mean that time and expertise won't be required.

Recent previous discussion at: https://news.ycombinator.com/item?id=26672009 .

Free software was always intended as Libre software or Freedom software, though.

The main concern is autonomy, not economic costs.

I expect you know this already, which is why the puppy analogy sort of fails.

Very fair point, since "free puppy" misses the "free as in freedom" aspect. I think it works better than than "free as in beer", which captures part of the CapEx dimension but none of the OpEx dimension.

I would argue yes and no here. If those are two hrs where you are not employed making $100, then its $55. If you have to give up 2 hrs of employed time to maintain this, then yes $255.

I love my free time and there is precious little. But I don't think of it as costing ME $100/hr when I wash, dry, and detail my car, especially as I like doing it.

It’s for people with enough spare time. If you find yourself having to cut your work hours to the tune of $1000/mo (or more realistically, personal time that you value) to self-host some stuff compared to the time it would take to maintain non-self-hosted equivalents, then by all means, don’t – but that’s definitely not how “actual cost” works otherwise.

And two hours a month seems high.

i understand what you were trying to say. you are right that people underestimate the work that goes into handling everything themselves.

You also need backups and cloud redundancy , if something would happen.

You'll need a devops team as soon as you use the cloud ( eg. kubernetes)

Those will cost easily more.

Reply on news.ycombinator.com