All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Hetzner vs OVH
I've been using Hetzner for around 16 years now (back when a lot of their pages were still lacking English translations) and currently I have 18 servers there (mostly in Falkenstein). I've got cheaper servers elsewhere, but Hetzner is where I've put machines where I need to be reasonably confident they will stay up, and that any faults will be dealt with quickly. So far I've been very happy in that regard, and have machines sitting around with over 5 years uptime. A few hard drives have died over the years (especially on the cheaper machines), but with software raid 1 on all servers, I'm yet to lose any data during a drive failure. Network failures can happen from time to time, but are a lot rarer now than they were over 10 years ago, and are usually resolved very quickly.
However the looming price increases are concerning. I haven't been told yet what the new pricing will be for my current servers come January 1st, but the closest equivalent for our higher end servers on the current pricing suggests costs for the more expensive of our servers will be going to be going from 190 euro to 275 euro/month which is a fairly hefty increase, and is starting to make rethink whether Hetzner is the right hosting provider for my budget now.
I've never looked that closely at OVH before, as it seemed they were more expensive, and people spoke less than kindly of their support (I don't need more support than an ability to trigger reboots/reinstalls when required , and get prompt responses when hardware dies - I'm fine doing all the actual operating system administration without their help).
However OVH are now cheaper for at least some of the use cases I have, and given I can get servers relatively close to the current hetzner boxes (in terms of ping times), I could probably do a slow transition to OVH over time.
So that leads me to my main question. Those who have used both OVH and Hetzner, how did you find their support compared (given my fairly low level of support requirements).
One of my concerns with Hetzner's latest pricing is that I'm not convinced that it is only about power price jumps (I'm sure some of it is), and that Hetzner may be using the power prices as a way to transition to a higher cost market segment, and may not care so much about trying to continue to take big chunks of the low end segment of the market.
I'd really prefer to stay, as overall I've been more than happy with Hetzner, but I fear being priced out, as our margins are not so fat that we can easily afford to absorb the upcoming price rises that will be kicking in on Jan 1st for all existing servers.
Are there other providers that I should be considering other than OVH (the closer to Falkenstein the better)?
Comments
Hetzner has better support, more bandwidth as standard, cheaper storage, customisable hardware and better physical datacenter standards
OVH has server-grade components (Hetzner cuts corners), automatic hardware replacement (no ticket required so less need for being on call overnight), lower prices at the low end, possibly better ddos protection
I ended up moving my Hetzner service to OVH after the first set of price rises and no issues at all. Drive replacement was less quick than I’m used to at Hetzner (had to wait till Monday vs reply within an hour), but this is comparing to Kimsufi which is not exactly known for good support
I would better suggest you move to @Zare or @PHP_Friends
Great support and network
. You might work out some custom things indeed.
I've been with OVH since a long time, and happy with the experience. However, I have to recognize I've been very lucky, and avoided all big problems OVH faced the last few years. (e.g.: the fire in Strasbourg DC), and in my luck, I never (yet) had an issue involving contacting the Support.
So happy experience in my case, but I can easily imagine the nightmare it can be, when craps happen.
You can reboot your server, and reinstall it anytime you want, and by yourself. As for hardware failure, I didn't yet face this problem, but OVH monitors the server, and a ticket is automatically opened, when the server is down. A tech checks the server, within an hour. Again, I didn't experience this myself, so I don't know if this is true, or not.
We are happy to offer you something, especially with the quantity we can give more discount than it is the case with our standard servers. In general we offer a support from 9-24 o'clock daily, beyond that you can reach our emergency support 24/7 by phone (English or German). Our location is directly in Frankfurt (InterXion FRA16 as well as maincubes FRA01).
Only benefits of OVH are the cheaper IPv4 pricing and good Anti-DDOS. Most comparable configs at OVH cost nearly twice as much as to those at Hetzner.
While OVH's incident support is good, their ticketing support is one of the worst in the industry.
You can expect a price increase by the end of 2023 at most hosts as they will most likely be signing new power contracts for 2024 at the end of the year, which could be higher if the power crisis is not resolved in Europe.
Thanks everyone for your replies.
After reading the comment on pricing from @Smith42 I went back and checked, because I thought I'd seen considerably better deals for some configurations from OVH, but it seems my comparison was somewhat of an apples vs oranges comparison, and yes, in general Hetzner does still look cheaper if you try to match up the exact hardware offering. However if you're willing to compromise slightly, OVH can provide opportunities for cheaper servers if you don't care as much about some aspects of the hardware.
For example, looking at a 512GB RAM PX93 on Hetzner (not a configuration we are currently using, we only have 256GB RAM in our most expensive server at the moment, but something we're looking to upgrade to soon), was going to be 453 Euro/month with two 1.9TB NVMe drives. With a zero commitment purchase OVH can do the same RAM and storage for only 381 Euro/Month (dropping to 343/Month with a 24 month commitment, something Hetzner doesn't seem to offer). However, the CPU at OVH is a AMD Epyc 7371 vs a Xeon W-2295 at Hetzner, and the Xeon is quite a bit better in single core performance (although similar for multi-core, despite less cores for the AMD). The OVH also has half the speed on the network port (500mb vs 1gb).
Hetzner has great (I'd say probably unbeatable for the level of service and reliability provided) base prices, but the main issue with 512GB pricing at hezner is that they can only do that with 64GB simms and because the base model pricing includes 32GB simms that you don't get a discount back on when you don't use them when upgrading to 512GB, the 512GB is priced at 242 euro over the base price (with a massive jump in per GB cost once you cross the 256GB barrier). OVH on the other hand are pricing 512GB at only 118 euro/month, and the upgrade to 1.92TB NVMe is also slightly cheaper at OVH (47 Euro at OVH vs 67 at Hetzner).
Given the bandwidth and CPU differences are not a deal breaker for us, it still seems the OVH deal is a bit more attractive all things considered (especially as we probably wouldn't mind committing to 24 months). This is a Rise-4 BTW, hosted in Frankfurt in case anyone was wondering.
I also like the fact that OVH will auto-raise tickets. At hetzner we do our own monitoring, and usually Hetzner will not be aware of a hardware failure until we tell them. This works ok, given they are pretty quick to respond, but if someone (usually me) sleeps through a mobile phone alert (or forget to take their phone off silent) something can be down for longer than we'd like.
At this point I'm considering getting a single OVH machine for testing, and see how things go for a reasonable period, this gives us time to see if we like their setup (although only one machine isn't a great test - the chance of needing support when you only have one machine vs 18 at a single provider isn't high). I probably wouldn't be interested in going with smaller boutique providers, the good ones usually have excellent service when available, but can sometimes struggle with the 24 hour thing.
Has anyone ever had any luck negotiating volume discounts with Hetzner. If so , what do they consider 'Volume', we have quite a few other non-Hetzner servers we could move if it helped with pricing (we could more than double the 18 we currently have there, although I suspect they can't touch the pricing we get on most of those outside Hetzner, so it would have to be a big volume discount to make any sense for us to move servers over to them).
How does OVH know if a server is shutdown by the customer or has an issue?
looking into logs ?
They have some automation around pinging the server. If server is not responding to pings automation try's to power it up. If it fails couple times then tech is being called.
You can disable the monitoring / auto hardware investigation through the panel if you are doing work that will stop it responding to ping
They don't by default. The other week I shutdown my kimsufi box accidentally instead of a local machine (
sudo haltin the wrong terminal window, I don't have that box much locked down against fuckups) and work up to emails to say they'd detected an unexpected shutdown and powered it back on.In theory they could automatically know the difference, but that would require a monitoring agent installed that could log externally “this was an orderly shutdown, not a more unexpected outage”.
Not that it would have made a difference in the above situation, but I either didn't know or had forgotten that. Useful reminder!
Question. How do you apply kernel updates, etc with 5 years uptime?
I can only think of one truly essential kernel bug fix in the last 5 years, which was a buffer overflow in the TCP fragmentation code. Other than that, kernel patches aren't all that essential as long as your userland is up-to-date, and if you have a very small attack window (e.g. only HTTP(S) and well audited server code), you might not even bother doing that.
On those machines we don't. But the machines with that kind of uptime are usually not providing services directly to end users, so are setup to not have ports open to the wider internet, and in that configuration I don't know of any priv elevation vulnerabilities in the currently run Kernel that can be exploited without local access (or that are not mitigatable without a new kernel with other measures such as firewall rules). There are some potential DoS exposures, but we are generally not a target for that type of thing, especially on these non-public facing boxes, and worst case, we get brought down by a DoS and have to reboot with an updated kernel. We haven't experienced that yet.
On other machines, we have to have publicly open services available, and those do have more regular reboots due to the larger vulnerability surface area those more open services provide.