Hetzner introduces new server product with insane parameters for 59 EUR/mth
hetzner.de
hetzner.de
I compared them some time ago to EC2 pricewise:
http://codemonkeyism.com/dark-side-virtualized-servers-cloud...
And setup of a new server is quite fast too.
Example: To have remote KVM/IP you'd need to first pay 15EUR for their flexipack for the privilege of adding additional features. then 19EUR mo and 149 EUR setup for KVM/IP Now you're up to $122USD/mo + almost $400USD for setup.
Compare to Incero's WebHostingTalk offer: http://www.webhostingtalk.com/showthread.php?t=1121998 which is $99/mo paid in quarterly batches. Which is server grade, better CPU, ECC RAM, with KVM/IP and is in the US (Texas).
Hetzner is still cheaper overall with the 32GB -- because they're not dealing with much more expensive server class ECC RAM, but I just feel its important to bring the initial low number you see: 59 to a realistic place...
I have been using an older server offer of theirs for years now, and never had to pay for anything other than the said fee.
Edit: 59.00/1.19 = 49.58 (it really is minus 16%, my old accounting teacher would kill me ;-)
I'd recommend Linode if you need really good service (they once tracked down and patched a kernel bug for me), but if you don't want amazing support and just want some really good specs at a good price Hetzner is great, especially if you're in Europe.
Intel® Core™ i7-2600 Quadcore incl. Hyper-Threading Technology
RAM 32 GB DDR3 RAM
Hard disks 2 x 3 TB SATA 6 Gb/s HDD
7200 rpm (Software-RAID 1)
NIC 1 GBit OnBoard
connected at 100 MBit
Backup Space 100 GB
Traffic Unlimited*TL;DR; Yes they're trustworthy, generally speaking
Of course, if you're working on a larger project with enough budget, services like Akamai provide methods of routing your static content via their network, further reducing the problem.
Edit: I see we might be exceptions. We have been using OVH since 2005, running up to 40 servers about 2 years ago. We are also running at iWeb, Softlayer, Amazon and Leaseweb. Hardware wise the only one we have issues with is OVH, and i'm talking every 2 weeks since 2005 (in 2006 they managed to lose all our data on 2 clustered RAID machines; one died, all corrupted, then when we were busy setting up the replacement, the other one died...). But they are so cheap and easy (like the easy OS installation which is cloud-like, already for many years) we stick with them for some projects which need high spec hardware in clusters. Note: the SSD drives are great, no problems with those so far (knock on wood).
They also do remarkably well: out of the thousands I've had from Leaseweb and OVH, OVH wins hands down.
I sort of wish hosting companies would publish the hardware failure rates -as seen by them- too - though I don't know how much of a competitive advantage that would be, if any.
I'm kind of wondering now if we are running different workloads; we are running very high IO and network stuff. Maybe we just hammer a lot harder on the disk than others are? With HP DL360s @ leaseweb we are running the same loads without any problem (since 2006...); 1 broken disk that ended up not actually being broken.
And yes the "host" is a very big datacenter, I have one server with them for 3 years, with pretty good uptime. 2-3 outages on their end over last 3 years. You need to remember that you need to buy additional packages for hardware raid or failover that will cost you about 20-30 EUR - if you need them, still it's one of best offers out there and def. best in Europe price wise.
You must also know that the EU paid some good money to get the datacenters there as part of economical development initiatives. Hetzner is operating since 1997.
Incredible, maybe they made a mistake on their site. Hetzner has a very good reputation here in Germany, otherwise i would expect something fishy.
Sadly I just reserved an instance for one year with Amazon the other month but this looks quite a good offer.
And you can run VMware esxi on these offerings.
Also I've been given faulty hardware at Hetzner and the support was quite ass-hatty. I said fuck it, replace the drive and reflash the OS, but they no, the RAID must rebuild. At the end of day, MySQL was still crashing under a heavy load, but that could have been software problem, not sure if I can blame Hetzner. What I didn't like, if you open the contract eg on 15th, then cancel a month ahead, eg on 10th, they will still bill you till the end of the month, 15 more days. They're still one of the best non cloud hosts though.
edit: now I remember - they actually insisted on the faulty drive (that I reported based on mdstat logs) to be hardware-checked in the machine while the machine is offline. Mind you, this drive was faulty from the get go.
We just ended up purchasing a new box, migrating over and letting them take the old one offline. We ended up paying less per month for twice the specs...
Deleted comment
One thing I have wondered about is whether hosting in Germany is a good idea.
First is the latency issues. Will serving web pages to customers on the west coast of the US really suffer? (this is in the minimum viable product stage-- we're not google yet)
Secondly, I'm not all that familiar with all the laws of the EU. I don't want to run afoul of them, but I haven't found a good summary of best practices for services hosted in the EU.
Less likely, but because it's Germany, if you're hosting bigoted content, particularly Nazi-associated content, that will be a problem.
Censoring is illegal from our constitution, so you can only get sued by a court, but when you take it down it usually would be dismissed. It is not even an important public topic here.
Anyway, I'm not doing anything controversial, and I won't even be hosting user generated content (at least not in the early days, though we might later).
Twitter took this as an example to justify their new censoring features, that annoyed me a little. There is a difference between censoring and enforcement trough court. For the second one you just need a system for takedown requests.
I recently tried to research what ECC means operationally (I'm not really an operations guy.) Is ECC good for correcting bits flipped by cosmic rays? Or is it going to tell you that memory is failing? I've never, in the past 20 years, had memory fail on me over time.
Also, I'm not going to be running a single server such that if a server goes down my service goes offline.
I'm running a Riak cluster, of at least 3 nodes, maybe 4 to start. If any server goes down, the service stays online.
My software architecture is, near as I understand it, well suited to commodity hardware.
The 10 euros extra per month, on 4 servers is 40 euros, which is nearly enough for an extra server.
So, wouldn't it be better to have 5 commodity servers than 4 servers with ECC?
I wouldn't know what to do if I started getting ECC errors anyway (I know, I'm stupid, but we're a startup, no budget to hire an ops person.) I'd probably just ask them to give me a new machine.
Why should you care? Well, if you are lucky, the machine crashes. If you are unlucky, it keeps running, and that bad data gets written to disk, or causes a cumulative error in some calculation, and you never know about it.
With the size of memories these days, chances of a corrupted memory are not exactly a rare occurrence.
I don't know enough about Riak, but a cluster is only going to help if key data structures are themselves checked for consistency across the cluster.