I think I'm done with low end providers

vitobotta

I have used mostly Hetzner for many years (and a little UpCloud and DigitalOcean, but mostly Hetzner), and I have never had any outage with them of any kind. Actually maybe once there were some speed issues with the network due to a temporary maintenance but not a big deal.

Over the years I have tried other, cheaper providers often out of curiosity to see if I could find something good enough that could help me save some cash. Every single time I went back to Hetzner due to technical issues.

I had promised myself not to bother anymore.... until I saw a pretty good offer here recently, so I migrated some stuff because I am trying to cut costs. First impressions were great, as well as the performance overall.

Now I'm 9 days in, and already got the first multi hour outage due to issues with the physical node.....

I know, node issues can happen anywhere..... but it doesn't look promising after just a few days.

Now I'll see if it gets resolved in a reasonable amount of time and then will see if I want to migrate back to Hetzner or stay and see if I have issues soon again.

Looks like I am not particularly lucky with low end providers

james50a

Honestly probably the smart decision if you really need high uptime. You get what you pay for essentially.

bikegremlin

That works as intended.

Out of shear sportsmanship, to not put the high-end providers completely out of business, the low-end providers stage regular downtimes, late support ticket replies and once a month they get completely wasted while sorting the tickets, just for a good measure.

The world is better that way.

angstrom

@vitobotta said:
I have used mostly Hetzner for many years (and a little UpCloud and DigitalOcean, but mostly Hetzner), and I have never had any outage with them of any kind. Actually maybe once there were some speed issues with the network due to a temporary maintenance but not a big deal.

Over the years I have tried other, cheaper providers often out of curiosity to see if I could find something good enough that could help me save some cash. Every single time I went back to Hetzner due to technical issues.

I had promised myself not to bother anymore.... until I saw a pretty good offer here recently, so I migrated some stuff because I am trying to cut costs. First impressions were great, as well as the performance overall.

Now I'm 9 days in, and already got the first multi hour outage due to issues with the physical node.....

I know, node issues can happen anywhere..... but it doesn't look promising after just a few days.

Now I'll see if it gets resolved in a reasonable amount of time and then will see if I want to migrate back to Hetzner or stay and see if I have issues soon again.

Looks like I am not particularly lucky with low end providers

Perhaps a little dramatic to start the same thread both here and there about this revelation ...

In any case, yes, most low-end providers won't be able to have better uptime than Hetzner Cloud over an extended period of time, but this is as expected. Nevertheless, there are a number of low-end providers that have pretty good uptime, and one chooses such a low-end provider for other reasons (recurring discounts, a desire to support smaller providers, etc.).

willie

I thought Hetzner was also a low end provider so at the beginning of your post I thought you were about to say you were moving away from them, heh.

Yeah I've had terrific hardware uptime and hardware support with Hetzner. There has been some trouble with network routes on and off though: I've had pretty big slowdowns getting stuff to and from North America.

OVH hardware has also worked very well, flash sale kimsufi pricing has been great, their network may be better than Hetzner's, but their support is notoriously terrible.

havoc

Don't buy frozen pizza and expect a gourmet meal

vitobotta

@angstrom said:

@vitobotta said:

Perhaps a little dramatic to start the same thread both here and there about this revelation ...

MichaelCee

You’ll be back. 😁

vish

@MichaelCee said:
You’ll be back. 😁

nullroute

A provider, whether low cost or not, should not imply the quality of its services provided in terms of fair use and quality. It's just a matter of time to find a balanced provider for your needs.

chimichurri

Four things:

• First of all, could you please post the name of the provider? Or at least the server location (ideally, the city, not just the jurisdiction) so that we could exclude at least some of the usual suspects..;)
• " I have never had any outage with them of any kind. " - sorry if I sound too skeptical, but, are you sure?... How are you monitoring your servers, is it by any chance something similar to what I'm doing, i.e. one minute interval health checks from multiple locations where failures are counted in case of 50%+1 checks failed (c/o Hetrix Tools)? Because, believe me, there is a difference between setting a 1m and a 5m interval, not to mention anything longer, as well as having a single vs. multiple check locations...
• How long was (or is?) the "multi hour outage" exactly, and what kind of SLA do you expect for Hetzner-level priced services, for the record? For reference, it seems their SLO is 99.9%, whereas https://uptime.is/99.95 i.e. some 4hrs/year is more or less the uptime (as per the Hetrix Tools setup described above) I'm getting from most of my low end storage servers, as well as my low end shared hosting. I also got some free Oracle Cloud VMs, they seem to have a 100% uptime, but they are not publicly available, so I can't vouch for their network availability using Hetrix Tools data in a similar manner.
• Pretty sure that hardware failures happen to Hetzner VPS too And since AFAIK they already phased out CEPH based cloud servers, I believe the only way to get ~100% uptime is to have some kind of underlying redundancy (which I believe is not available with their NVMe based services) - you could either do it yourself by using e.g. k8s or in some cases, leave it up to the provider, e.g., I believe Upcloud who you've mentioned in the OP have redundant hardware with live migration (this, BTW, might also partially explain why they are more expensive than Hetzner Cloud...)

angstrom

@chimichurri said: First of all, could you please post the name of the provider? Or at least the server location (ideally, the city, not just the jurisdiction) so that we could exclude at least some of the usual suspects..;)

First of all, this thread is a dramatic overreaction by the OP.

Second, the low-end provider in question very probably just had an unfortunate incident, so there's no need to name and shame the provider as though the provider systematically had such incidents.

CamoYoshi

Why put the failure on the hosts when you can instead make your... whatever high availability across multiple nodes, multiple companies, multiple datacenters, multiple regions? It doesn't excuse the outages but it will make it a much more trivial occurrence when they go offline.

rcy026

My personal experience is quite the opposite.

I use high-end hosting for a lot of my customers simply because they demand it. I also have several personal projects where I get the cheapest hosting I can get since it's mostly for fun and testing anyway.
I have an automated monitoring infrastructure so I basically monitor everything, and looking at the stats for the past 3 years several of the ~$4 a month providers equals or even outperforms many of the high-end providers, including AWS and Azure.

That being said, no, you should not host critical services at lowend providers. If you try to save a few bucks by moving your "critical" service to another provider, then sorry, but your service is not critical.