I'm a recent Hetzner customer.

32Gb/i7.

I used them for 5 weeks, in that five week period I had 5 1-2 day blocks where the machine was unreachable.

>A incoming attack influence the network for few minutes. The network department has solved the issue.

few minutes was 31 hours

our network-team told us, that there was again an attack which caused the issues you mentioned.

27 hours

currently we can't see any issue, is this still given?

connection timed out (no idea how long as I'd not installed monitoring at that point).

etc etc.

I canceled shortly after.

I'm back on Linode for now (I actually changed the design of the product I'm developing so that I can store files onto S3 so I can stay with Linode and move easily in the future).

A fast server is a paperweight without reliable network access.

Just as a counter point, I'm both a Linode and Hetzner user and am happy with both. My Hetzner server has had solid service for about 18 months now (with a couple of brief restarts in there) and certainly no prolonged outages.

I am not particularly keen to move entirely to them, however, as I just have a gut feeling I'd prefer to have the service of someone like Linode in a true emergency. As a high powered server for experiments, side projects or things that don't demand 100% uptime (e.g. my company's collaboration software), however, Hetzner has proven superb so far.

There was a true emergency with Linode.

They've been hacked badly twice and both times withheld information from their customers. Bizarre that you would prefer this type of service than from others.

i host a few sites with moderate traffic on an ex10 server. I can't remember of any network issues. Current uptime:

12:13:14 up 183 days, 21:23, 2 users, load average: 0.51, 0.55, 0.49

The problem is unreliability of hardware. When one of the hard disks broke in the previous server, it was faster to setup a new server to transfer everything rather than waiting for the repair.

I concur with the unreliability of hardware. Back in January a production Hetzer box of mine saw both HDDs die at the same time. That somewhat killed the redundancy benefit of RAID.

That said, it did spur a month-long effort to truly revamp my dusty backup-and-restore process so I'm in a much better position for when it happens again.

You're telling me swapping HDDs and letting RAID rebuild is slower than setting up the server again?!

A failed HDD should NEVER, EVER, bring a server down.

i should have mentioned that both disks reporting errors. The DNS change required less downtime than the downtime of backing up everything and setting up the servers from the start.

I had a Hetzner server and was very very satisfied both by the performance and the price. Does anyone know a comparable offering in the US, for people who want servers there? I think I heard OVH?

Yes OVH should give you the same sort of things, and they now have an american datacenter (you can chose which datacenter you want your servers to be in when ordering)

Thank you, the prices look comparable. Does anyone have experience with them? How are they?

OVH are great as long as you don't need to get in touch with them - If one of their servers break, you're better migrating to a new box than trying to fix it.

Same experience here. This is why we always have more servers up than we need to. One time it took us almost a week of mailing before they replaced a broken hard drive. But they have to save money somewhere.

I'm strongly advising against cheep providers like ovh or hetzner without proper redundancy.

Is there a Hetzner equivalent in the US or Canada?

I've been keeping my eyes peeled for one and.. there are actually several. http://netdepot.com/ is the one I've come closest to ordering from so far but I haven't yet done so. I like that they do "buy downs" where you can pay for extra hardware up front though. It's not quite as cheap as Hetzner but it's not miles away.

Others on my radar are https://hivelocity.net and http://www.100tb.com/ - at the VPS level, http://buyvm.net/ and http://prgmr.com/xen/ are both cheap for the quality you appear to get (but I can't yet vouch for this directly).

The "buydown" option provides a nice alternative between colocation and dedicated server. With initial setup fee of $1300, you can get a server with similar specs to m2.2xlarge + 4000 IOPS for just $139 a month, while not having to worry about replacing failed hardware parts and resource contention. Sweet.

Thanks for the links. Not sure if any of them fit the bill for me, sadly - I need at least one box with large RAIDed SSDs for a database. Hetzner seems to offer this.

OVH has very similar prices and you can peep at their http://status.ovh.com/ to see how they have been going with uptime.

Yeah, I've been very tempted, but I've unfortunately seen some horror stories about them that have given me pause. Have you given them a try? If so, would you run a business on them?

We have nearly 100 servers in the Montreal BHS datacentre. Been very happy so far. The web GUI is fairly buggy and they have no automatic billing so you have to manually renew each server but they tell me this will be fixed soon.

Good enough for me, thanks. Though not being able to pull together something basic like automatic billing is pretty worrying :-/

We're happy with OVH (we have some boxes in Montreal and some in their France location) - good prices, no issues after a year+ usage.

OVH actually does NOT have very similar prices. Hetzner is by far the cheapest dedicated hosting that I know of.

So, I have to be cool with having terrible reliability, no durability, no services I rely on (SQS, S3, EBS), increased latency for my users (our userbase is 99% north american), and extended downtimes anytime anything goes wrong with hardware, networking, etc? Maybe there's a reason for the price difference?

Dude, I thought you kept everything on a single "big iron" server? Why are you using AWS?