This website made me reconsider if 99.9999% SLA and always online is really that important for lots of websites. I used to self-host my own server in my own apartment, but gave up on that after my internet provider had their first major outage that lasted almost a week. Also I didn’t want to worry about the fire hazard, or the server going down when I was traveling etc.
But a solar powered server seems interesting, if the batteries could be made fire safe.
This was cool 5 years ago, now you can get solar panels and a home battery big enough to power this thing 24x7 without an issue, even if the sun was fully blacked out for a week
edit: a 40kwh battery would last with the power draw:
Power usage 2.62W
like over a year
There’s a lot of articles about supply based usage vs demand based energy production on that website. The idea that it turns off every now and then is sort of part of the authors’ intention of showing “we can live with some outages”.
This article in particular is a good one: https://solar.lowtechmagazine.com/2017/09/how-not-to-run-a-modern-society-on-solar-and-wind-power-alone/
100% uptime since the last time we turned it on!
Use a gell cell lead acid battery (the kind that go in regular UPSs). No fire hazzard, ans bonus, they can survive freezing temperatures just fine if outdoors. Downside is replacing a cheap battery every 3 years.
Appreciate the website itself, with it’s dithered low-colour PNG illustrations.
The uptime looks better than GitHubs…
I know this is a typo, but I’m sad Guthub isn’t a food delivery company.
gut pull sandwich.com
gut commit -m “ate too much”
gut push
garage is a software project made for communal home hosting. Tech literates will talk about its S3 API but that’s the least interesting part: the most interesting is that it has been purposely built to be hosted on (actual) commodity hardware like second-hand nucs, running on flaky home power and internet and linked together through usual isps. Nodes may come up and down but the cluster withstands all of that.
Garage would be a pretty nice project to run a solar-powered mini-cluster. If sufficiently far from each other, you could even run without polluting batteries !
That’s really neat
This actually cool.
All websites go offline when there is no power …
Companies claiming 99.9999 uptime is an absurd goal for a SLA, and wouldn’t even survive a small internet outage like a DDoS attack or Akamai outage. I remember in the 90s selling the idea of 5 9’s - 99.999% on my resume until some manager who Interviewed me asked me to state the expected downtime. I stopped claiming that number because it’s impossible without full active redundancy in different locations and not dependent on external providers like cloudflare ever going down.
And please don’t argue maintenance windows “cover it” because that’s not how they work in the real world today.
Downtime per year:
- 99% ~3.65 days (5,256 min)
- 99.9% ~8.77 hours (526 min)
- 99.99% ~52.6 min
- 99.999% ~5.26 min
- 99.9999% ~31.5 seconds
For my home lab I’m happy my stuff is online and content I hardly lose access when there’s a power outage thanks to my ups. But if I tracked I’d be closer to 95%. Good enough but I can’t wait till I just don’t care to run anything.
I worked for a “99.999%” up time facility. No one talks about the exceptions that don’t effect that number. Some are out of our hands like weather or a knob w/ a backhoe.
Others are just made up by upper management to hit the numbers for their bonuses.
I have never been a fan of anything above 3 nines of uptime even in a business situation.
Your business will not collapse because of 10 min of downtime a week, and if it does that is not technologies fault.
I don’t even bother to track my home setup.
I found the GitHub employee! 😉
There are plentiful scenarios where 10 minutes of sequential downtime would be a critical issue. Say you’re a telecoms or internet provider - even a one minute delay to a life-saving message can be enough for someone to die.
Furthermore, lots of things are interconnected now. Your business might be able to handle 10 minutes of downtime, but lets say my business relies on 12 different service providers like you, who each have 10 minutes of downtime during a week using your rationale. That could up being 2hrs of downtime each week.
99.999% is about 5 minutes and 16 seconds of downtime per year.
12 services down perfectly sequentially works out to 76 minutes per year, not 2hours per week.
Not saying there aren’t applications where that can’t be catastrophic, but you’re scale is off for three 9 downtime.
Yes, there are some that do, I should have said that. But the number of orgs that need extreme uptime is so small compared to business that think they do need it but really don’t.
My home server is 100% up, because I never check it and that’s how it works, right?
I didn’t turn it down, so it’s obviously up.
Schrödingers home server
I don’t even bother to track my home setup
I host home assistant and if anything goes down I’ll notice pretty fast.
There’s a nifty project on Mastodon where someone is running an instance entirely on solar (and I think a battery?). They’ve been tinkering with some of the code to do things like optimise images for reduced weight or CPU cost.
The project is called @solaradmin@solarcene.community and the account for the person running it is on a separate instance, presumably so people can message them about problems: @smallsolar@techhub.social.
congrats guys! we killed it!
Uptime for home users is useless. More important is recovery, reboot and resume (the 3’Rs)
You server should reboot automatically, storage should restore itself from unsafe shutdown and your services should resume operations like nothing happened.
This means test your reboot, ensure stuff doesn’t break, and so on.
Failing storage after an hard reboot is the hardest.
For network, i have two ISP connections and a an auto switch script Incase one goes down. But honestly it’s overkill and not needed for 99% of home hosters.
Buying some kind of UPS like a battery powered power strip is also nice but keep in mind that require maintenance to replace batteries once every few years (long enough to forget) and so be useless. The ups shall only last enough to perform a controlled server shutdown using NUTS.
Current uptime of over 23 weeks, not bad
It’s late summer now, best possible time to measure solar-powered uptime. Roughly 5 months, i.e. since mid March, i.e. springtime in Spain. Many daylight hours.
deleted by creator
Lead-acid batteries work if you don’t care about weight and lithium-iron phosphate if you do
You’re better off with the LiFePO4 batteries even if you don’t care about weight. A lead acid battery will wear out quickly when you cycle it daily. Lithium also has a higher charging efficiency, so you get more usable power from the solar panel.
Lithium-iron is about 2x more expensive, there might be different availability and energy use during manufacture is probably higher
