251 comments

[ 3.7 ms ] story [ 370 ms ] thread
Note these are atom based.

But I bet nginx could still crank out static files from them.

Translation: http://translate.google.com/translate?sl=auto&tl=en&js=n&pre...

    We have 3 servers in the KS series:

    KS-2G ATOM powered with 2G of RAM, 500GB,  100Mbps 2.99e/mois.

    KS-4G 4GB of RAM, 2x500GB software raid 9.99e/mois 

    KS-16G Core i5 with 16G, VT and 2x1TB 19.99e/mois

3 euros is $4 USD and 20 euros is $26.50 USD

They are also on the UK site:

http://www.ovh.co.uk/dedicated_servers/kimsufi.xml

That $26.50 server is currently $40 in the US (CA)

http://www.ovh.com/us/dedicated-servers/kimsufi2.xml

The real deal here IMHO is the i5 for $26.50, that cpu can run laps around the atom and can run in "turbo" mode near 3.5ghz all day.

They're also great for backups.
Even if it can't do serious web hosting of any kind, that's 500GB of storage for $4/month. Great solution for offsite backups, bearing in mind of course that there's no redundancy.
They used to have unmetered bandwidth, so you could also use them for big transfer.
Pretty certain, like most hosts, they won't allow the storage to be used for remote backups. The 500GB is usually only for data/files being served/processed by the server.
With an encrypted volume, how on earth would they know?
These are physical, dedicated servers. You get full control over what's going on.
It's dedicated hosting, not some shared managed thing. You do whatever you want with it.
Thats a dedicated server, you are confusing it with shared hosting.
(comment deleted)
I've been using a 1TB one for over a year as a backup, nothing was against this in the contract. Mind you, there's a 5TB per month traffic cap, after which the bandwidth is limited from 100Mbps down to 10Mbps.
(comment deleted)
e = € but this it before taxes. After taxes: 3.58€ 11.95€ and 23.91€ (= 31.61)
They remove tax if you're outside of France. I paid 2.99€/month (but then Paypal added their conversion fees, I paid 12,58$ CAD for 3 months.
For the benefit of those of us who don't know much about Atom, could you explain what the limitations are of using Atom for webserving?
It is a very slow CPU designed for power efficiency, not speed.

As such, it can be quite surprisingly slow, managing to limit a lot of loads that most devs take for granted these days.

The Atom is Intel's weakest chip. It is designed for super-low power consumption, which means that they cost very little money to run. (so usually, data-centers like running them, because they don't use much electricity. Naturally, the savings are typically passed onto the consumer).

The Atom is Intel's line of chips designed to compete against the iPad. It is still fully x86 compatible, its just slow. Roughly... the speed of a Pentium 4 or so.

Current generation Atoms are 32-bit only, which means 4GB maximum RAM (not an issue for smaller webservers with only 2GB).

Atoms tend to be best for I/O limited tasks. If you do video streaming for instance, the vast majority of your CPU power is going to be "wasted" on waiting for the Hard Drives and the Network. So the Atom makes a good, cheap CPU for that kind of task.

For tasks with tons of computations, (ie: game servers), the Atom is woefully inadequate.

You can purchase an Atom computer for ~$165: http://www.superbiiz.com/detail.php?name=MB-ZID60-U. So play around with the performance yourself, and see if its enough.

Atoms predate iPads by approximately 11 billion years. They're the chips used in Windows-based netbooks, nettops etc.
> Current generation Atoms are 32-bit only

Some lines of Atoms since mid-2010 or so have supported x86_64. The Sxxxx "server" atoms released at the end of 2012 are all 64-bit-enabled, for example, although they can still only take 8GB of RAM: http://ark.intel.com/products/series/71265

I agree that this is mainly interesting if you're I/O-limited. Most benchmarks show a "regular" recent-gen Intel CPU performing around 5x-10x the Atom, so if you're CPU-limited, even a low-end VPS that gives you an average of 1/4 share of a core will probably outperform sole tenancy on an Atom core. But a high-contention VPS in most cases will give you much worse I/O performance. Also, the storage space and bandwidth quota here are much better than with a comparably priced VPS.

Thanks for pointing that out. It seems odd to me that a system that only supports single-channel DDR2 RAM would be 64-bit however. With so little RAM, the benefits of 64-bit computing are basically void.
But it is more work to make a 32 bit system. They are probably all 64 bit locked into 32 bit compatibility mode.
Their 64bit kernels can run a 32bit distro too. With multiarch you also get to cherry-pick packages of the other arch.
The virtual space can be useful too. I remember reading that Go had some issues due to the limited address space in 32-bit systems.
Yeah, that's an issue with conservative garbage collectors. When memory actually in use is a large fraction of the address space, then random values can look like pointers too often. When valid_memory << address_space, on the other hand, most non-pointer values won't alias a valid pointer.

The specific issue with Go was significantly mitigated in Go 1.1 by making much more of the GC precise rather than conservative. Here's a comment from about a month ago on the current state: http://code.google.com/p/go/issues/detail?id=909#c59

Isn't the x86_64 instruction set more RISC-like and compiler-friendly (i.e. more registers, 3-address instructions)?
> The Atom is Intel's line of chips designed to compete against the iPad

The Atom came out in 2008, two years before the iPad, and come from a lineage started in 2003. It was designed for the now dead netbook category that required low power consumption.

64-bit versions have been available since 2010, both versions used by OVH are x86-64. There are also server-specific versions with support for ECC memory, though they doesn't seem to be using them.

iPad's CPU/GPU uses ~5W of power. These Atoms use ~8W of power.

Regardless of the 'ordering', the Atom line of processors are designed for the <10W form factor. True, Atoms existed before the iPad, but they are in fact designed to compete in that power-range. The analogy works very well: the iPad 4th generation A6x processor is just slightly slower than the newest Clovertrail Atoms. The older Atoms (like the 330) probably are a bit slower than an iPad.

http://www.anandtech.com/show/6522/the-clover-trail-atom-z27...

So yes, perhaps I misspoke earlier. Nonetheless, I stand behind the analogy. Intel Atoms are Intel's version of the iPad A6x. They use roughly the same power and give roughly the same performance... but Atoms give you full x86 compatibility.

Perhaps a more appropriate way to say it... is that the Atom is designed to be a competitor to low-power ARM chips in general. Intel is making strides in making extremely slow, but power efficient chips. And in Performance/Watt, they're roughly on the same scale.

Yes, meaning they were made to compete with ARM makes more sense, not the ipad (non-existent att) specifically. It was Intel's bet on mobile. The cortex processors used in the iOS line have only caught up recently regarding performance.
The Atom was made for the burgeoning market of low power, low performance, low priced netbooks. That was seen as the probable breakout market, though it was short-lived.

So Intel wanted a fairly power efficient chip, but they also wanted to ensure that it didn't cannibalize their own sales, so they intentionally crippled it from a performance perspective, not least by always building it on the last or second-last process. This is an aspect of the Atom that is missed by so many, sure that Intel was caught with their pants down by ARM: Intel's biggest fear wasn't ARM, but that their pricey high-end CPUs would get replaced by their low cost variants. They still fight with this paranoia to this day.

The Atom was neither inspired by the iPad (obviously given that it far preceded it) or even ARM. At best you could say Intel had some concern about Transmeta, leading to some of the early Atom work.

This isn't a minor error of "ordering". These were critical mistakes in your retelling of history.

I'm well aware of that fact, and I am also aware of the fact that BayTrail Atoms are correcting this mistake. (Finally, 22nm Atoms for realz this time.)

I have admitted my mistake with an addendum, although I cannot edit my earlier post anymore. What more do you want?

50 pushups and a $10 contribution to the EFF.
(comment deleted)
> The Atom was made for the burgeoning market of low power, low performance, low priced netbooks.

This is incorrect. The Atom CPU was initially meant for smartphones, but the first models way overshooted the available power envelope. Netbooks were created by Asus when Intel offered a lot of cheap, slow chips for sale that were good for little else, and they created a new, shortlived segment. No-one at Intel thought about netbooks when designing Atom.

Didn't we already cover this? It is revisionist fiction, built around the notion that Intel was so outclassed by ARM, when really Intel was just short-sighted about how important smartphones would become.

The first Atom processors actually had extremely competitive power characteristics. The only supporting chipset, however, was a) a power pig, because Intel just didn't bother despite it being a much easier task than making a processor (I have a little file server running a dual-core Atom, the processor being air cooled while the chipset has a fan. As a humorous aside, altogether that micro-PC uses about the same power as my new i7 Mac Mini under standard loads), b) completely built for traditional PC uses. Do you need IDE on your smartphone, for instance? How about an A20 line?

The "glut of inventory" notion is not realistic. It doesn't work that way.

Intel had no intentions in the smartphone industry at the time. If they did, not only would they have actually made a decent chipset (they happen to be the most advanced chipmaker in the world, with the world's best fabs), they would have made a prototype (which is what all manufacturers do when they try to enter a market - a proof of concept). There were absolutely none until later Atom variants.

People forget it now, but it was initiatives like One Laptop Per Child that initiated the netbook craze -- that program talking about ultra-inexpensive laptops, leading to a lot of people saying "You know...I too would like an inexpensive laptop that I could just bang around and leave in the car and..." (Slashdot was full of people trying to repurpose such devices), and with that an industry was born.

Yup, OLPC was what "woke up" Intel, making them begin to build Atoms. Frankly, Intel had the foresight to start the Atom project to hedge their bets a little bit.

They never "really" tried to compete in the smartphone market (which is ~2W), the Atom was instead always targeted at ~10W. (IIRC, Intel's real focus at that time were Consumer Ultra-Low Voltage, CULV processors at the 18W envelope. Today... known as Ultrabooks). However, as Smartphones became the next hot thing, it is now obvious that Intel has to scale the Atom down even further. And unfortunately, the netbook market imploded.

Its not all wasted however, IMO, netbooks "became" tablets. From a CPU perspective, there is little difference between Netbooks and Tablets. You want low-power consumption between 5W to 10W, and lower costs. In fact, as netbooks were "dying out", they slowly became tablets. See AsusTab Smart for example, it basically is a Netbook without a keyboard, but built with a touchscreen. (same price range as Netbooks of old, but just in tablet form).

Anyway, Intel hasn't forgotten the Smartphone market either. Thus the Atoms that worked closer to the 2W SoC benchmark... Medfield, Clovertrail, and soon to be Bay Trail Atoms. Clovertrail is winning a few benchmarks here and there, and Bay Trail will be built on Intel's latest process for once. So it looks like Intel is finally taking the smartphone market seriously. But only the last generation or so are actual attempts at a Smartphone chip... and even then, there isn't a single Atom out there that is built on Intel's latest 22nm process. That will only come from the future Bay Trail Atoms.

> The first Atom processors actually had extremely competitive power characteristics. The only supporting chipset, however, was a) a power pig, because Intel just didn't bother despite it being a much easier task than making a processor

To add some numbers:

The most popular first generation Atom, the N270, had a TDP of 2.5 W. The accompanying chipset and I/O controllers however had a combined TDP of 9.3 W.

Source: http://en.wikipedia.org/wiki/Bonnell_%28microarchitecture%29...

They are on the level of a Pentium 4, but running at 1.6GHz. Thats bad news for a Pentium 4, and the Atoms...
If you web application could run ok (for the number of visitors) in a netbook with mechanical disk, then an Atom is ok.

Note that I've seen people running gnome on a netbook.

Still, I see this, and I think more about: "backups", "configuration management", "shell access", "VPN to other more powerful dedicated servers in OVH", "personal storage", "scheduled tasks", "monitoring systems", "status pages", etc...

The price is nothing, I still ask myself if it's rentable because "it's an Atom" and changes the power bill, or this is more a unprofitable product to attract you to other products.

Lets say 3€ * 12m = 36€ year, It takes some years to recover the hardware costs (without count electricity bills, company salaries, connectivity, taxes, etc).

> The price is nothing, I still ask myself if it's rentable because "it's an Atom" and changes the power bill, or this is more a unprofitable product to attract you to other products.

pretty sure it's the latter. Even if there is minimal human effort on part of OVH involved that probably leads to unprofitability. For example a RAM stick breaks and has to be switched out, I guess it takes a systems engineer in the NOC at least 30 minutes to locate the server, disconnect it, replace the stick, mount the server again, make sure it is running fine. Assuming that employee get's paid 20 GBP, that is a 10 pound cost plus replacement RAM which is...I dont know...let's say 20 GBP, recouping that kind of money would take about a year (I know very dirty napkin calculations).

I bet they have a handful of these gimmick servers and they are going to sell out very soon. Comparable to the Raspberry Pi colocation edis offers ( http://www.edis.at/en/server/colocation/austria/raspberrypi/ ) try to order and you will find that they are sold out.

>For example a RAM stick breaks and has to be switched out, I guess it takes a systems engineer in the NOC at least 30 minutes to locate the server, disconnect it, replace the stick, mount the server

These servers run without cases, using water cooling. Hardware failure automatically notifies the NOC with the location of the failed hardware and replacement parts are already waiting on the desk.

>handful of these gimmick servers

OVH is one of the largest ISPs in the world, and is well known for disruptive pricing and services, such as 100/100 Mbps broadband to the home for $20/month.

Don't get me wrong, this is an awesome offer but because I believe that these may be not available for a long time I am currently in the check out process, ordering one of these for a year even though I don't have an immediate use.

Edit: Order placed, eagerly awaiting my Isgenug (that is what they are called for the German market) :)

"is enough"?
yep translates to Isenough only the german "is" is a colloquial version of "ist" which is proper spelling.
"kimsufi" is actually phonetic in French for "qui me suffit", ie that suffices me, that is enough for me. I didn't know they translated those.
The reason this works is because only a few would face hardware issues, so while there would be a few unprofitable accounts as a result of hardware issues, the majority would be just fine. It's the same reason stores accept returns and don't go broke because of it; only a few customers actually return a purchase.
The other responses to this have been woefully speculative at best, plain wrong in others. The short answer is to rent one for a month and benchmark your particular application.

I was surprised to discover that with a workload composed entirely of XML parsing using Python+lxml, an OVH 1.8ghz Atom N2800 beat hands down my mid-2010 2.4ghz Macbook Pro. This isn't a crazy result either. For instance, while the Macbook is equipped with dual cores, each core supports only one hardware thread, and thus is blocked while waiting to complete a load from RAM (a common occurrence when dealing with giant cache-unfriendly DOM trees that don't benefit from the Core 2's much increased L2).

In contrast the N2800 has 2 cores with 2 hardware threads each (hyperthreading), so each core can progress while its buddy thread is stalled. In my particular workload this was enough to beat by a significant margin a much faster, expensive and power-hungry processor.

No offense... I don't buy it.

Core2s have significantly better memory bandwidth, because they're dual-channel. They stream memory significantly more efficiently than N2800 Atoms. When a Core2 Duo has a memory stall, it can start executing other instructions out of order to compensate. Typically, out-of-order cores seem to do more for efficiency than hyperthreading.

If Apple screwed up by only giving a single DIMM (it happens on lower-end hardware, but I'd be pissed if I learned that an expensive Macbook were improperly configured memory-wise)... maybe the Atom would have a chance. But a properly-configured dual-channel RAM Core2 Duo, as old as it is, still would stream data faster from RAM than an Atom could. Almost twice as fast in fact.

If the N2800 does perform better on this workload, I'd be definitely interested in seeing the exact numbers... if you don't mind.

I actually do own a Hyperthreaded Atom (N570) and Core2 Duo (some old crap desktop version) myself. So... I'd be able to repeat the experiment :-p

All the bandwidth in the world shouldn't matter if the underlying technology has the same latency and the memory bus is serving a fixed number of small requests in parallel, right? Overall processing speed in an uncached memory bound application is limited by the latency of the storage tech (somewhere around 10 million "requests"/sec assuming 100ns access time, and assuming the app did absolutely nothing except randomly access memory).

In my case I had 1 software thread per thread, I'm not sure if over-subscription would make any difference. Also, out of ordering processing is only possible in the absence of data dependencies (I have no idea how to even go about measuring this). Also tree structures are no different to random walks from the hardware's perspective, i.e. the hardware has no useful predictive ability to prefetch data in this case.

The main point was that it's so cheap to test, there's little value in speculating about architecture specifics few are qualified to understand, assuming they're publicly documented in the first place

Fair enough. But given my experience with Atoms... the potential of hitting CPU bottlenecks is definitely higher on Atoms than on other computers. The significantly smaller cache and lower clock certainly don't help either.

You're right in that its extremely easy to test, even to buy yourself an atom at home. Netbooks are regularly under $300, Clovertrail Atom Tablets are ~$400. Older Atom 330-based Netbooks are probably even cheaper. Its not like you're trying to test out the performance of a 64-core Quad Opteron or something.

And of course, the ability to just rent one for 3 months at a time from Kimsufi for less than $20 is always available.

The Atom range are intended to be low power draw CPUs for use in things like netbooks and server environments where I/O and bandwidth are going to be the bottleneck not CPU.

As such their suitability depends greatly on your particular application.

They aren't so good for high concurrency for instance, but if you are using a mainly event driven server where concurrency isn't a massive issue you'll not notice that limitation.

At this price, the most cost effective option is going to be to rant one for a month or few and give it a try. In fact at that price if the machine works for you it is an absolute bargain, you could pay more for a much smaller VPS where you are competing with others for I/O bandwidth. I have a machine at OVH currently for a few bits and bobs - for my use pattern I might be better off cancelling it and getting a few of these instead for the same price.

Wow! I'm taken down my VMs this weekend.
Hmmmmmmmmm Debian (the sensible choice) or Arch (the slightly more interesting one)?
Go for Arch! Enjoy pacman and be on the bleeding edge. Added bonus: new performance improvements will hit your tiny server faster, thus: more speed :)!
Arch is definitely a viable option for a personal server.
I have run Arch for two years on a tiny personal server. It is fun but depending on how you use it and what Alan is currently breaking, it can be a lot of fiddling.

Just as one example, sysvinit - systemd migration took some careful file editing not to botch things remotely. At one point, merging all binaries in /usr/bin incentivized me to move from the no longer supported GRUB-legacy to GRUB2. GRUB2 installer failed leaving a zombie process that prevented anything from touching the mbr...

I wiped it, installed Wheezy and I can tell you that it requires far less maintenance and you can still pull bits and pieces from testing if you really want to.

By the price of two coffees you get a server connected to internet.

When I started with this, the price of a server was the salary of an adult of medium class working during one year (and you got it without connection to internet, neither hosting facilities)...

I wish the price of technology at home (devices, connections, etc), could go down the same way for everybody.

Where do you get two coffees for less than €2.99?!
I can't speak for the OP, but where I live in Spain it's not uncommon for a standard coffee to be about... €1.2 to €1.5.
Here in Portugal, a €0.70 coffee is somewhat expensive ;)

That said, it should be noted that here in the Iberian Peninsula we just mean coffee, not buckets of water, cream, sugar and milk with some coffee mixed in, like those served at Starbucks.

Go to a good English greasy spoon rather than a posh coffee shop.
Those places usually sell you a cup of cheap supermarket instant coffee...
Here in Portugal a cup coffee usually costs between 0.50€ and 1€
(comment deleted)
It was just a symbolic reference.

In Spain (where I did grow) as well as in The Netherlands (where I work now) you can.

Of course, in both places, you can get more expensive ones too. Depends on the shop.

Including the taxes, it works out to a bit under $5US. A Starbucks coffee is around $2.
Taxes are already included. Prices are never given without taxes in Europe except in B2B where your customers don't pay VAT anyway.
From the linked page:

"All prices exclude VAT"

In Italy of course! An espresso at your favourite bar[1] ranges between 0.80€ and 1.00€[2]

[1] expensive exceptions apply, like in really touristic spots like Piazza San Marco in Venezia or similar [2] of course not served to the table but at the counter (or bench or stand or whatever you call it in english) like we do typically:)

Any bandwidth included? If not, how much is transfer?
100 mbit up until 3 TB, later 10 mbit cap.
Note traffic within the OVH network isn't counted.

For certain applications (which you may or may not know about) that's quite significant.

But these are Atom based meaning they are not great for a part of those certain applications :p
Ooh, ooh, I know, you mean BitTorrent seeding.
Indeed. For the uninitiated, I would venture to guess a majority of OVH network traffic is BitTorrent.
Actually they cracked down on their seedbox clients a few months ago. Found they were using a disproportionate amount of bandwidth.
The general terms of conditions explicitly mention seedbox's.

[7.2] "The bandwidth is no longer guaranteed when the server or servers are used for the following activities: ... Server(s) used for downloading and sending files on peer to peer networks (including but not limited to: seedbox)"

Quite the unnecessary rule I believe. 100 Mbps = 3.2TB in a month.
Quite necessary - 100 x 3600 x 24 x 30/8 = 32TB / month.
Your math is wrong by an order of magnitude, it's ~32TB.
Actually I think they removed the bandwidth cap earlier this week. So 100 mbit unmetered for 2.99 €/month. Crazy.
I wonder what price is shown to US, UK and French customers?

German customers have to pay €3.99 for KS2G. The German price should be 1 cent cheaper, and not 1 Euro more expensive, as French VAT is 19.6% while German VAT is 19%.

See http://www.ovh.de/dedicated_server/isgenug.xml if this is possible outside Germany. You can not see ovh.com inside Germany, as ovh.com is redirected to ovh.de.

The french site http://www.kimsufi.com/fr/ (not sure it's accessible outside France) shows the new prices. It seems the german site hasn't been updated yet.
ovh.ie (Ireland) shows €2.99 for KS2G with VAT at 23%.
On the german site the prices are shown including taxes, on the others excluding taxes. It's minor differences from rounding to "nice" numbers.
Where are these located? I'm wondering how bad latency from the U.S. would be...
Roubaix (North of France). Try pinging productism.com and report the ping from where you are?

Data center list: http://www.ovh.co.uk/dedicated_servers/data_centre_selection...

112ms from Michigan, USA.
Sounds high. Should be closer to 60-70ms from the East Coast.
From Metro Boston: round-trip min/avg/max/stddev = 100.297/104.900/110.846/3.406 ms
From New York on a major corporate network: round-trip min/avg/max/stddev = 75.940/79.012/81.289/1.950 ms

That's similar to the ping times to my current VPS on the West Coast.

Well, Michigan isn't exactly "east coast".
Not bad. 50ms from me (south England), only 5ms slower than pinging my servers located in the midlands.
Sounds a bit high as well. I'm pinging London-based servers from my OVH servers in less than 25ms
May vary based on last-mile stuff. I'm getting 55ms to you from my residential DSL in Copenhagen, but I get 20ms if I ping from a university server, also in Copenhagen.

Surprisingly constant 148ms from a VPS on the west-coast US.

From a server in the midlands I'm getting a consistent 8.3ms
You can also use http://proof.ovh.net/ but the GRA datacenter, where these servers are located, does not have a speed/ping test available yet.
Are you sure they are located in the newly opened GRA data center? If so, my bad.
These are all in GRA. I get 10ms from London.

  11 packets transmitted, 11 packets received, 0.0% packet loss
  round-trip min/avg/max/stddev = 111.052/116.042/120.117/3.412 ms
Columbus, Ohio
Looks like you get "Western Europe Datacenter (GRA)". And it ends up being $13.74 USD for 3 months.
In France, near Paris. Ping strims.pl to get an idea.
(comment deleted)
9 packets transmitted, 9 packets received, 0.0% packet loss round-trip min/avg/max/stddev = 39.689/40.127/40.869/0.371 ms

From Brussels

Amazing for the disk space.

Anyone recommends the company? Never heard of.

In my opinion they are number one in France. Their European network is nice too. The US one is rather new, I have no idea but be warned that theses new servers are not available at their American (Beauharnois, Canada) datacenter.
"Since 2012 the company has been classed as the largest server host in the world"

https://en.wikipedia.org/wiki/Ovh

Don't know if the Wikipedia is correct, but I know them (I'm in EU).

It's correct, in terms of physical servers installed. That said, there's no reliable comparison (that I know of) with VPS instances hosted by well known US companies (such as Amazon) because they are not quite as willing to release figures.
While it wouldn't cover VPC clients you can get a max figure by looking at the number of ip addresses allocated to AWS. They list them all here: https://forums.aws.amazon.com/ann.jspa?annID=1701

I think the last time I did the math it worked out to something like 2M addresses in US-EAST.

OVH is a very large dedicated hosting company. They now have a nice location in North America (Montreal) as well, with tremendous prices. They generally get solid reviews. If I had to peg them, it'd be a B rating or 8 out of 10. Weak customer service, but you can pay a modest fee and get immediate attention. Their hardware is solid.

Whether you use them or not is probably down to three issues: do you need managed servers and rapid response from a tech inside the company; if you need US based hosting, they don't offer that, but the Montreal location gets great speeds to most of the US; the listed kimsufi servers are obviously located in Europe.

They actually had a very serious security breach few days ago. But I would still recommend them over some of the others.
My short experience was very good.

A few VPSs for low traffic sites and a big dedicated for a drupal.

Everything was fine, no surprises, and they have a soap API which allows you to automate everything from billing, to dns, machine creation/resize/destroy, etc.

Nice.

My only caveat was about their custom kernels, but that is something usual in lot of providers.

Is there a way of using your own custom distro or image; vs the ones they provide?
When I used OVH about two years ago, on dedicated servers, it was possible. And I think still will be in those.
Keep in mind that's not RAID disk, so if that hard drive goes, so does all your data.
RAID isn't backup anyway -- it's for uptime.
That VAT costs more than the service. Hard to believe.

  > £2.49 excl. VAT /month
  > ( £2.99 VAT inc)
That's £2.69 service and £0.50 VAT. Seems reasonable to me.
How can VAT cost more than the service itself, when VAT is set at 20% of the cost?

(I believe its £2.49 without VAT, £2.99 with VAT included)

> That VAT costs more than the service.

I suspect the error here is confusing the VAT inclusive cost with the cost of the VAT alone.

Try ovh.ie. You have to send proof of address, but you don't have to pay VAT if you live in the US.
Another interesting use may be for an always-on BitTorrent Sync node. I have a Raspberry Pi for this purpose, but the 500 GB of storage space is much better than the 16 GB SD card I currently have.
You can always attach an external usb drive to your PI. I built a Time Capsule drive with my Pi, works flawlessly through network storage.
Do you have a link with more info on this setup?
I've never tried it, but something like this should work:

http://www.raspberrypi.org/phpBB3/viewtopic.php?f=36&t=47029

It's a similar process to setting up Netatalk on any Ubuntu server to get Apple computers to recognize a Linux server as a Time Machine endpoint.

It should be noted that the Pi is really, really slow. Too slow for this sort of application, IMHO.

An actual Time Capsule can be annoyingly slow sometimes, and I would expect it to be about five to ten times as fast as a Pi-based solution.

The link you posted has some good advice though. Like not using HFS+. Apart from HFS+ arguably being an antiquated filesystem, the Linux driver is in a horrible state and will cause corruption when used in rw mode. ext4 is the sensible choice.

Same here, I have a Pi has an rsync remote for my git-annex repository, storing on an USB drive. The speed isn't impressive, but it works quite well.
Anyone try this yet? Do they include any sort of remote console access?
I don't think it's there. Kimsufi is a low cost alternative to normal OVH services. I have few kimsufi servers for over year now and I never had to use remote console there.
Just to add, I have found that really only options that are available are: * hardware reboot * os reinstalation

But there is a option called "netboot" which looks like it might be what you are looking for, but unfortunately I can't try it now on any of my servers.

Netboot enables you to boot in a rescue environment. From there, you can usually mount your disk, do some basic tests, and fix mistakes.

Not as efficient as a direct console access, but for the occasional "Oh , I just firewalled myself" moment, it's enough.

Thanks; so basically if you bork your box your options are (1) netboot, and failing that (2) reinstall the OS. Reasonable enough, for the price. I suppose that this is why we have Chef and Puppet
I once firewalled myself and asked for help, a technician re-opened port 22 for me after a few hours (you can get instant support if you pay but it was for dev so I didn't care).
I'm running Minecraft servers on a KS-8G (seems to be discontinued now - basically 3.2Ghz i3, 8GB RAM, 1TB HDD)

* Their management interface is garbage, especially compared to sites like Linode. Pay your bills, restart the server, cpu graphs, set up rDNS and that's about it.

* Customer support response times tend to be ~1day or more.

* The network has been down for a period of over an hour twice this year.

However, for the value for money in terms of hardware, I don't mind overlooking the above, especially considering it's just a game server.

For anything more important, you might want to consider other providers, though I'm sure customer support is better on the higher end packages which come with phone support and so on.

Are you running vanilla Minecraft or a FTB-like modpack? How many users can you comfortably support?
Both. One vanilla server and one FTB Ultimate. Occasionally there's a second vanilla server running for playing maps as a group. The FTB server starts to have mostly cpu usage related issues with 10 people online, especially if some of them are in the Twilight Forest but it's only for a small group of people so that only happens if almost everyone is online. I also had lag when generating chunks, but then I put the world on a ramdisk which helps somewhat, and I have a plugin to back it up every hour.

Vanilla has had 14 people without issue. It's CPU usage is minimal compared to the FTB pack. Could probably go to 20 or 30 with no issue.

I just tried to order and it only let me choose from one country (for my personal address) and that is UK. No option to select USA.
You have to use their french website to select other countries: http://www.ovh.com/fr/index.xml
They removed the US as an option
No, it's there, follow this link, order something, it will allow you to create new contact with United States as your country: https://www.ovh.com/fr/commande/kimsufi.cgi?hard=134sk1 At this point it doesn't matter what to order, you will have to place your order again once VAT is removed from your account.
Just did this. Now (in french of course) they want me to send an email to verify my identity with a scan of my drivers license or passport before being able to purchase. Not sure if I want to do that...
I am using the US version of their dedicated server to run URL unshortening API service http://api.unshort.me/api.html . The web stack is Nginx + Python (Flask) + Couchbase. So far, I haven't notice any issue with their server.
You mean Canadia
DC photos can be found at https://lafibre.info/datacenter/data-center-ovh-roubaix-1/. It doesn't get any cheaper than that.

I'm guessing they buy mini-itx boards from China by the container load.

This is their data center in Roubaix. The datacenter these new dedicated servers are available in is in their new datacenter, Graveline (Graveline France).

Probably similar spec, though.

Aye. But if you are paying less than $75/month for a dedicated server, what do you expect? XD
This is very exciting, but appears to be UK only. The only country available on the selection is UK and it wants a UK address.
This is for every kimsufi offer, by default kimsufi.com redirects me to http://www.kimsufi.com/fr/ but you have to add different coutry TLD to see prices in different currencies.
You need to go to the appropriate OVH site for your country. OVH.es for Spain, OVH.pl for Poland, etc.
Kimusufi.com did not have this offer and kimsufi.us is something else entirely, but kimsufi.ie seems to accept Americans.
Are they using ECC or non-ECC RAM in the server? I'm guessing the latter, can't find evidence either way.
What are the advantages/disadvantages of either?
ECC Ram has a checksum that can correct one wrong bit in memory and detect two wrong bits in memory.
Is there a way to get the 2.99 euro price from the US? How about from Canada?
order with OVH.IE
the price is actually in pounds, not in euro. (still cheap however)
Hmm, that's all nice, but what about securing my data against snooping? I'm wondering specifically how I would prevent people from snooping data while in transit (MITM) and prevent someone with physical access to the hard drive from reading my bits? To simplify the question, assume I just want to store and retrieve plain files. Bonus points if I can also safely share them in some way (i.e. share one file without also giving up security on the rest).
For backup, you could use duplicity or obnam, both of which gpg encrypt incremental backups.

For file storage, you can use git-annex which gpg encrypts data stored on rsync special remotes. I hope to add gpg encrypted git repositories to it soon too.

Or tahoe-lafs, which encrypts files and allows sharing them by sharing a special url.

What? Encrypt it on your system, store the encrypted blob there. Add a simple web server to host them.

If you do any crypto on these servers, obviously anyone with physical access could compromise it (although potentially not without alerting you that it had been).

No doubt the new rates for their dedicated servers are in response to hetzner's new rates. ISPs seem to have this unfortunate ability to retain ridiculous profit margins indefinitely.

I guess the tactic is quite simple: Be extortionate while you have market share. Undercut your opponents when it starts being threatened.

Not that I'm complaining when stuff like this happens, I like OVH, sometimes.

Anybody have one of these that can benchmark it against a $5 Digital Ocean VPS?

$2.99 is a crazy price for a dedi.

It's 2.99 euro, or $3.97 (not $2.99). Still under $5, though.
would this be a good fit for storing and serving lots of images?
Yes. In fact that'd be a perfect use for this given the low-powered CPU. Just make sure you have decent backups and replication to protect against node failure.
Additional information: You can't add ip failover on your server. i ordered one 3€ server
Why would you want to host your stuff on a company who's CEO isn't developer-friendly?
Because it is budget friendly. Sad, I know. But true.
Not developer-friendly??
He's talking about his tweets on how he prohibits his devs from posting on Github: https://news.ycombinator.com/item?id=6064087
I see.

That's totally unrelated to their service though. Lots of companies have an exclusivity policy.

Nothing to do with whatever development happens in their servers done by their customers.

OVH is much less expensive than Amazon EC2... Someone said in a HN discussion earlier this week that he was not able to find providers significantly cheaper than Amazon. Well OVH proves this is possible.

This OVH dedicated server is comparable to an Amazon EC2 Standard small (m1.small) instance. Taking an Amazon reserved instance, "heavy utilization" (to minimize cost), and in their less expensive region, with a 1-year term, the upfront payment is $169 plus $0.014 per hour, which amounts to $24.30 per month.

Compare this to OVH is only 3€, that is $4.00 per month... 1/6th the price! And OVH has 500GB storage vs 160GB for Amazon!

>Someone said in a HN discussion earlier this week that he was not able to find providers significantly cheaper than Amazon

Yes but EC2 is a cloud provider and these are dedicated machines which are harder to scale without setting up a complex cluster.

You have API to order new server on OVH but it's not delivery instant
People don't use AWS because it's cheap. People use it because its scalable. You can deploy hundreds of new servers in minutes literally to meet any demand. You pay for that flexibility.
People use it because they think they need the scalability. The reality is that almost nobody that use EC2 have the ability to scale up or down that quickly, and even fewer needs it:

Most sites have a fairly stable base load, and most dedicated providers these day can provision in less than 24 hours, so you at most need the ability to handle short spikes or growth over hours rather than days.

And EC2 is expensive enough that you can handle huge variations by buying extra dedicated capacity to cope. But the (far) cheaper option still is to handle your base load on a cheaper provider, and build your system to be able to scale using EC2 or another cloud provider to handle spikes.

WIth more and more providers providing "hybrid" services with colocation, dedicated, dedicated vms (single vm per physical server), and cloud servers all (or most) on offer from one location, this gets even easier - you can sometimes even get virtual private lans across dedicated and cloud servers without having to mess with your own vpn or pay for the bandwidth