I'm not surprised. OVH deals are insanely good, often less than 10% of the cost of the equivalent AWS/azure/google compute costs in terms of the raw power you get.
I'm consistently surprised that more people don't use them for workloads that don't require super high reliability, but do use an enormous amount of compute/storage/bandwidth. I think a lot of services fall into this category.
they're not insanely good, they're just very close to the cost of the actual financed hardware.
AWS et al. charge insane markups because people fall for their marketing, which is cleverly disguised as spec sheets and technical documentation and other things engineers who think they are not susceptible to marketing fall easily for.
someone's paying for all those free AWS instances, and it ain't the low end customers.
edit: if you want to know the true price of the hardware, take the purchase price of your standard supermicro server, and multiply it by .025 - that's what big companies pay per month for hardware, usually on 3-year leases.
i.e. $10,000 box = costs $250/month. A decent cloud provider can usually get about $3000+ a month of revenue off of a box like that. even after you add power, networking, staff, it's still a large markup.
And, lets say you only need 3-5 of those $250/month instances... given AWS's services (or azure, or google) it isn't really a bad deal. I'm not saying it's for everyone. I'm still running some personal projects on smaller hosting providers... but with a modest budget, AWS and the like are decent offerings for what you get.
If you're designing for distributed data/services from the start, it's better still.
OVH prices are so amazing, I wonder if they must be running at a loss. I use them to run Aberoth (http://www.aberoth.com), and I always keep my eye out for a better deal, but nothing comes close for high RAM servers. They also seem to be continuously sold out: http://www.soyoustart.com/us/essential-servers.xml, so there is clearly a huge market for these cheap servers. I wonder why no one else seems to be able to compete close to their price point.
Been using their <$5/month Atom server for quite some time now. Zero problems, and I've had months pushing 500GiB of traffic, getting full 100Mbit pretty much all the time.
Their cheapest Atom seems to be $7? It also looks like the cheapest one they have in their Canadian data center is a $21 Core i3 (which is still a very good deal, with 2 cores/8 GB RAM/1 TB HDD/unlimited traffic).
There was an offer on within the last year or so that took them down to crazy low prices. It apparently caused a lot of customer churn which is why they now do setup fees :(
Never had a problem with their service, but their pricing is all over the place
The only issue I've had with OVH and Kimsufi servers was that any incoming or outgoing SSH connections would also terminate on the first connection attempt. Always. Mac, Linux, Windows - it didn't matter. I've never had this issue with any other host.
I got so annoyed with it that I created a simple bash script that would attempt to connect, pipe the results to /dev/null and then connect again. Always failed on the first attempt, always worked the second. Their support weren't able to help me either.
The only upside was that it forced me to add decent connection error-handling to my scripts.
Having the full cloud ecosystem on AWS/google is great. Launch a server that store files in S3, talks to a DynamoDB and memcache server and has a CDN and load balancer. The amount of back-end work required to make that is minimal and coordinating all those services (scaling, configuration, redundency) is so easy you don't even need a sys admin to do it.
You need a way to declare your cluster using some orchestration tool like ansible / CloudFormation. Once you have that it's easy to setup/teardown a whole cluster.
To further separate dev/prod you can create separate VPC (virtual networks) and use IAM (ACL) to limit access. It's also possible to create an AWS account for each developer and keep an eye on the spending with consolidated billing but in practice devs often want to share infrastructure for example when testing out something together.
Downtime sucks, if it happens a lot try investing in a better Internet connection. It's possible to re-use ansible formulas to deploy something locally but usually you'll end up booting machines that themselves need internet to fetch packages / dependencies anyways.
I can't give a very complete answer as I've been using AWS to work on a side project for about a whole week.
I try to design my app with 12-factor in mind (http://12factor.net/) so all the configuration is passed to the app at run time (either ENV vars, config files, args or pulled from etcd). I'm trying to run the exact same code while in dev and while deployed in prod. For example I will absolutely try and avoid having "if Env == Test" or "if Env == Prod" in my code.
For example, my service connects to S3 to store/retrieve files. The access key, secret key and even the bucket name are all passed as ENV variables so I can work on a test bucket easily with a test api key without changing code. Some services I can run locally (memcache or redis are really easy to spin up locally) and some services I'll rely on Amazon to provide like S3 even in dev/test. Of course this requires a reliable internet connection but it's the best way to deal with this for me as a single developer/devops.
I'm sure bigger teams handling larger projects could give a lot of valuable information on the subject, I would love to hear about it too.
I've been using them with no issues for the past 3 months or so. I priced out AWS just for kicks and it would end up costing us around 15x more from the $174/month at OVH. At least for me, the cloud has a long way to go before prices become even remotely comparable to dedicated hardware.
The math on AWS makes sense if you either have very unpredictable workloads or are a startup that lacks capital to get into any medium to long term commitment.
I see the latter case used as an argument lots but I just don't seem to understand it.
OVH will rent you servers for as little as a month at a time, for a _massive_ reduction in cost over AWS. We're talking >10x. Are there really startups that only have capital to last a week?
There is some back and forth on the "best" approach to hosting infrastructure, but there is no silver bullet or always right answer. There is a full spectrum for hosting infrastructure options, and each company has to pick the hue/tier(s) that delivers for them best a particular point in time and realize that it will likely change as time moves on. At each hue/tier, you trade man-hours and in-house expertise for cost. Typically, you see companies move backwards down the choices as time goes on to find that optimal spot.
- Multi DC colocation, building your own servers/network equipment
- Multi DC colocation, buying servers/network equipment
- Single DC colocation, building your own servers/network equipment
- Single DC colocation, buying servers/network equipment
- Dedicated servers
- VPS servers
- Virtualized servers as a service with barebone supporting
infrastructure (e.g. Digital Ocean)
- Virtualized servers as a service with robust supporting
infrastructure (e.g. AWS)
- Platform as a service (e.g. Heroku)
You do see companies get into "trouble" where they might have picked the right hue/tier but haven't reflected upon the choice in some time. In those cases, you might see a startup dropping 60k a month on AWS services when they would have no earthly business spending more than 10k amortized on a multi-DC/added hires approach (I've observed this exact scenario).
> You do see companies get into "trouble" where they might have picked the right hue/tier but haven't reflected upon the choice in some time. In those cases, you might see a startup dropping 60k a month on AWS services when they would have no earthly business spending more than 10k amortized on a multi-DC/added hires approach (I've observed this exact scenario).
I too have seen startups burning six figures monthly in AWS because they're afraid of buying gear, renting colo space, and hiring sys|network|linux admins.
i run a hosting company. or - i guess 'cloud computing' is what they call it these days.
a huge number of our sales conversations are people who spend $50k+ on aws or something similar, but are scared shitless to move off of it because they don't know what they're doing and their trusted technical advisors got them into the situation they're in. you literally have to talk them down, like a kitten that climbed up a tree and can't get back down.
imagine amazon had your business by the balls to the tune of $1M/year and the cost is increasing faster than your business can sustain, and you have no idea what hosting really is, or how it actually works. this happens. a lot. there are a lot of very lucky people out there who build incredibly successful businesses with a trail of terrible decisions.
for what it's worth, we've figured out the threshold is basically $10k. if you're spending over 10k on amazon, you might just be doing it because it's convenient, not because you have any idea of what you're doing, either technically or from a P&L perspective.
Just moved to OVH bare metal for my primary servers. Deal is insanely good, as is performance. If you need to scale beyond commodity cloud or you need higher single-threaded performance without the hypervisor overhead it's definitely worth a look.
>> "They explained to us that a key part of OVH’s growth strategy is growing their US business, including establishing a data center on the west coast, and attracting more US customers with an offering they call Dedicated Cloud."
Runabove (one of OVH's brands) offers very compelling pricing for object storage and transfer. $0.01/GB for transfer. It seems very tempting to store to both AWS and Runabove, keeping AWS in backup in case Runabove has issues.
This is very interesting for the european market. I was downvoted for bashing Hetzner and other German based hosting companies when Amazon launched its Frankfurt datacenter. But at least OVH learned that they need to compete and reinvent their business based on dynamic/cloud services.
I wonder what happens next with the competitors: 1&1/United Internet is probably the biggest hosting company but they still don't show signs of a new cloud approach. As far as I can see they just focus on their "website builder" product which is something like Squarespace and probably the future of the former "shared hosting" segment. They tried to build an infrastructure as a service company called Profitbricks but it took them years to provide decent API adapters to popular cloud connection frameworks like fog.io or Apache's libcloud. There service is limited (couldn't find something like route53 or a global CDN-option like Cloudfront, Akamai)
Then there is still Hetzner which tries to compete by price. They still offer powerful dedicated servers but still lack any professional feature including an API to order or cancel new servers.
Let's see, maybe both are going to acquire capital in the future and start inventing again.
On the other Hand DigitalOcean with its Amsterdam and London based locations is very cheap, powerful and can already be use fully automated. I would probably build my SaaS business on their infrastracture and then scale out using AWS or Google.
To anyone who wants to see a damn cool modular datacenter implementation they have, and how your server fits into it, checkout this video of their newish DC up in Canada: https://www.youtube.com/watch?v=Fnarvq0XpkA#t=100
When you use said DC for your servers and you see the severing naming conventions and routing paths involved... its cool to kinda be able to visualize how it physically is laid out.
That's the risk you run with any service provider.
At one point I had servers at The Planet/Softlayer and after a year they had a new deal with the same server/hardware with more memory for half of what I was paying. I ended up ordering a new server, and having to move my data and everything because the sales rep couldn't get me the new deal on my old server.
I've tried, like when I needed a bigger server but everything was sold-out, so I asked to customer services if they can apply the new price for my old server, they refused, after this I waited for a 7 months for the new servers begin available again, and then in dec/2013 they also discontinued the "new" servers without anything else in place and I cancelled my account.
I'm consistently surprised that more people don't use them for workloads that don't require super high reliability, but do use an enormous amount of compute/storage/bandwidth. I think a lot of services fall into this category.