If I were to do some rough math I’d say it would cost about $300/CAD per month.
My goal is once we are approved to start accepting donations that I can purchase dedicated hardware for this instance. I’d get a used server at about $2300 which would be sufficient a good amount of extra users and through it into its own dedicated shared colo at about $100/month. Factor in about $300-400 a year for drive replacements and we are left with $2300 / 12 month= 191.66 + 100/month for the shared 1u colo + a budget of $400 for drive failures throughout the year $33/month.
191.66 + 100 + 33=$324.66/monthfor the first year dropping to about $133 per month after the first 12 months. It’s worth noting that this method would give us double the amount of resources and quite a bit of extra storage.Ideally we don’t keep this instance on a single server forever and start to think about spreading it over multiple hosts at or after around 100K users (or less if the number of active users is high).
If someone wanted to host an instance they would not need to allocate as much resources as I have to this instance and depending on how active the instance gets could run off something a lot less powerful.
This is exactly what I was after thanks for the good rundown! Also thanks for all the time money and effort spent on all of this
I really appreciate the information, it’s very interesting to me. Given that you have a fairly specific price in mind for a server, what kind of hardware are you thinking of?
Something with Dual CPUs, at least 128GB ram, dual 750W PSUs, hardware raid (12Gbps) and 8 x 2.5" SAS/SATA slots for SSD Drives on a raid 10
Ddr4 or 5? I might have 128gb of rdimm ddr4 I’m never going to use sitting on my desk.
ECC memory though?
Really nice breakdown thanks for sharing! Totally reasonable goals to reach too.
I love the transparency. I think we can easily reach that mark. Whenever you get approved for donations we’ll be ready. I’ve got at least tree fiddy in my account
Have you considered the implications of hardware failure on uptime? And where the cost to maintain a physical hardware will come from? What about scaling requirements?
I’m not a network engineer, but I’ve been involved in the corporate argument of Cloud vs On-Prem. hosting for years now. The costs always come out better for Cloud when factoring in other indirect costs like facilities and labor.
Granted it’s always been on the scale of hundreds of millions to billions of dollars, and I haven’t run the numbers on smaller requirements. I just wouldn’t want to expose additional points of failure in return for slightly lower monthly costs.
He mentioned colo, so it sounds like he’s already decided against on-prem.
What are you doing for backups?
That’s the “drive replacements” part I believe
Drive replacement != backup
Drive replacement = maintenance, a subset of physical security.
Backup = logical security.
The purpose of backup is to prevent loss of data in general, not only on account of drive failure, but also other sources such as malicious activity
There is a pinned post on this community. It reads:
Firstly, the issue of donations. Since the inception of this instance, your most frequent request has been the ability to make contributions to support my initiative. While initially, I had never intended to accept donations, I’ve come to realize the value this brings in ensuring our platform’s sustainability. In response to your requests, within the next week, I will be introducing several options for those of you who wish to donate. I want to emphasize that these donations are entirely optional and will directly support our instance’s operational necessities - dedicated hardware, colocation fees, email services, and more.
