601–610 of 643 posts
I finally started migrating our app from Heroku to Hetzner. Started by moving Postgres, and boy, what a difference in performance. I pay 1/5th of what I previously paid, and performance is through the roof. Like, as if we had rewritten our Rails app in Rust (metaphorically speaking). I know that in theory you get much more from managed postgres than a simple self-managed server, but I have pgbouncer and Wal-G set up,…
For now?
How much of AWS EC2 you can get out for 70 bucks?
Now, this OVH still makes money on it. They make money, despite these servers require actual human being to put them into datacenter, plug network, power, etc. You are literally getting the OG RAW POWER and a slice of a datacenter for 70 bucks and they still make money.
How much amazon makes on every single silly vm that they charge for compute, storage, network, ip, network again, oh and credits, cpu credits, startup credits, whatever credits, oh an api calls ;]
Earlier quoted context omitted.
Cogent just offered me this colo deal: Full Rack = $100/month* with $500 install, Power (20A) = $350/month with $500 install, DIA (1Gbps) = $300/month Total = $750/month plus $1,000 Install on 12 month term
Where is this, if I may ask?
Earlier quoted context omitted.
Sure, but that’s because Azure. I’m sorry someone made the decision to go there. AWS & GCP, stock outs at least used to be nearly unheard of.
Until you hit a certain scale. I totally agree about Azure being the worst of the three, they wanted us to commit to certain use before even buying hardware themselves. Crazy… But I also had capacity issues with Google at large scales in many zones.
Earlier quoted context omitted.
Until you hit a certain scale. I totally agree about Azure being the worst of the three, they wanted us to commit to certain use before even buying hardware themselves. Crazy… But I also had capacity issues with Google at large scales in many zones.
What sort of scale, if you don’t mind me asking?
One gameserver was 40vCPU and 256G of RAM, we had about 30-50 before we’d see some issues in some regions. (this is from memory now unfortunately).
Sao Paulo and Tokyo being the worst, but Singapore, Australia and Mumbai also had issues at various times.
The other places where we hit hard limits was Los Angeles, but we had more than a hundred instances then.
The issue with the hard limits is that it’ll be one zone thats exceeded and the API will fail, so you have to retry with another zone in the same region, but you don’t get to practice building your autoscaler before you actually need it.
Earlier quoted context omitted.
What sort of scale, if you don’t mind me asking?
Hey, sure! That’s important context. One gameserver was 40vCPU and 256G of RAM, we had about 30-50 before we’d see some issues in some regions. (this is from memory now unfortunately). Sao Paulo and Tokyo being the worst, but Singapore, Australia and Mumbai also had issues at various times. The other places where we hit hard limits was Los Angeles, but we had more than a hundred instances then. The issue with the har…
That is a lot lower than I expected, but I also imagine that’s a sizable order that they like getting.
Earlier quoted context omitted.
I have not run into this multiple times. You said that, not me. I said something different. Hetzner is the only cloud provider that cut me off. The other provider was not a cloud vendor. Even if it was a shared instance, people don't hire a 48 core server just to use 1 or 2 cores. It makes no sense to rent out a big shared server and then expect users to not use it. Someone would rent it out only if they have exhaust…
> I have not run into this multiple times. You said that, not me. I said something different. Hetzner is the only cloud provider that cut me off. The other provider was not a cloud vendor. So, you have been cut off multiple times. > Even if it was a shared instance, people don't hire a 48 core server just to use 1 or 2 cores. No, but that is also not what is happening. > Something tells me that your idea of computing…
Earlier quoted context omitted.
There was no crypto mining involved whatsoever. There was some crypto analysis involved, among unrelated other analyses, but no mining. If you think about it, Hetzner had to be spying on my activities in very close detail to see what I am doing. Such unnecessary spying (without a court order) alone should detract anyone from using them. Assuming they copied my disk image and subjected it to a scan, it's very possible…
Hetzner just like every major other server/cloud rental out their does prohibit crypto related activities in their products. This is not limited to "mining". They do not need to access you disk to determine that. They can just observe behavioural patterns and use heuristics. E.g. 24/7 100% cpu, low data, port traffic on typical crypto related protocols to known crypto endpoints will raise red flags. They can terminat…
The heuristics you note are of garbage quality, full of errors in classification, with a faulty classification being their default case. It's when someone is first found guilty of being a witch, then a dumb association is noted to label them as such.
If a user is using a shared CPU all of the time, then it's the job of a good cloud vendor to throttle the user when resources are needed for other users. Hetzner simply does not have this technical capability which goes to show how weak it is.
The sheer number of blatant lies in your comment go to show that it is you who has extremely bad faith all around. You are a bad person who has much to lose with Hetzner being exposed for what it is, a fraud.
Earlier quoted context omitted.
Yes, this is what we do for example for the tailscale operator - but it's tedious to convert yaml to tf - and more importantly: error prone to correctly adapt upstream changes to update deployments as upstream refine their helm/k8s yaml files.
Terraform/OpenTofu is good for infrastructure but it becomes a pain if you use it for k8s deployments. I suggest using GitOps (ArgoCD / flux) for everything inside the cluster and OpenTofu for the cluster itself.
Earlier quoted context omitted.
Hetzner only has one Datacenter/AZ per region. So you either risk a single region failure taking you down, or you lose performance from transferring data to another location.
These physics are exactly the same with AWS et al.