Ask HN: Why are cheaper hosting services like OVH not very popular?

I'm gonna say overly high expectations probably have an effect here. As you said, these things aren't necessary for a startup with a small amount of users, but... very few startups seem to realise that, or consider that they won't become the next Google or Facebook overnight.

So services like OVH aren't as popular as AWS because startups always buy on the assumption they'll suddenly hit it big, even if their needs are more modest.

It could also come down to branding too; AWS and Google Cloud and Azure have done a lot of marketing in order to get themselves out there in front of the startup crowd, with whole conferences and events on how to use them. Hence some people are likely using them because it's 'the thing startups do'.

I think it is because many people's bill is not primarily in EC2.

AWS, Google Cloud, and Azure's biggest value add come from managed services, like auto scaling databases and VMs. It makes it a lot easier for a small startup to build out their services without thinking about scaling challenges.

Snapchat, for example, was built on google app engine and to this day has billions of dollars in contracts with Google Cloud. They scaled their app to millions of users remarkably well with what I hear was just 2 guys.

No one does a startup because they expect it to stay small, so eating the upfront cost of using a managed service is a reasonable price to pay for peace of mind if things take off and the team can't keep up.

Hiring people to manage infrastructure and do it well is likely more expensive in many scenarios.

The big cloud providers all give a massive amount of credits for startups to use. Massive meaning multiple hundreds of thousands of dollars. So it’s easy for startups to hop on for the free credits and a couple years later be too busy on product cycles to make time for a move. The “lock-in” is strong.

I don't have answer to you question, just a few notes about another low cost proder - Hetzner: * They have RAID by default * They have private network included now (vSwitch)

Drawbacks from my expirience: * You have to monitor hardware by yourself (SMART, CPU and HDD temperature) and requests checks if you suspect hardware failure * Decreased flexibility (curretly Hetzner have limited cloud support, but it's very limited). Probably doesn't matter for most startups. * More manual work for administration (provided services are more low level)

AWS/GC/Azure have many different services well-known APIs/SKDs, which are like Lego to build products quickly. So if all you need is hosting for a webapp, you aren't taking full advantage of the offering. Then providers like DigitalOcean/Linode/OVH can definitely make sense. Heroku isn't quite like either of these, but another option if you want even less ops work.

Anyway, those legos also give you a certain future-proofing. Even with normal growth, it's easy to outgrow environments, and provider migrations/cross-provider apps are a pain.

There are all kinds of good and valid business reasons why, and all kinds of good and perhaps valid counterarguments to those reasons. None of which are the real answer. The real answer is that the CEO/CTO/CxO doesn't want to have to think about these things, or getting expertise in-house for these things. They want it to be like electricity or water; something they have to pay for, but don't have to think about. Whether or not that's what really happens is irrelevant; it's what they think will happen, and to be honest it is what often happens. Even if they're paying more $$, they paying less in mental bandwidth, and never have to worry about whether they have (retained or retrained) sufficient in-house talent to maintain it. That's the real reason.

OVH is not typically seen as a big provider in the US but has a nice euro footprint. So some of it is branding and marketing. I used to work with most of the American leadership at OVH and know for a fact that they view the next couple years as an opportunity to become one of the key cloud providers in the us (along with the big 3). They see themselves as every bit as technically strong, just need to market better.

their website is awful, when I last paid them they still didn't do recurring billing, so for each service (i.e. virtual machine) you have to remember pay them monthly for each VM, each month

or you pay annually... then if one of your VMs dies good luck getting their CS to do anything about it (or even ever finding out why)

Before moving to GCP we actually did experiment with putting some of our services on a VPS at OVH as well as other providers to see how well they would handle our traffic. The issue we had with OVH specifically was that whenever traffic got high they thought there was an attack in place and automatic firewalls went up that took us down for hours at a time. Upon contacting support we were told "After looking into the matter, it would seem that our VPS do not have a profile that would accomodate your traffic spikes"

We've never had an issue like this after moving to GCP.

Because EC2 is not the only service that you'll use. Having services such as SQS, RDS, EBS, and Route53 all available and integrated into the VPC and access controlled via IAM is incredibly valuable.

External tools such as Terraform and Kubernetes already supports AWS. For smaller hosting providers this is usually not the case.

Did you use spot instances when possible?

Did you use autoscaling to keep your instance count as small as possible?

These could be reasons why your costs were so high.

Another reason why companies use AWS is that it allows them to move faster. If you're afraid of lock-in and want to build everything yourself then yes, AWS will be damn expensive. But if you use the services they provide, you can build things pretty quickly without much workforce.

For me, it's how the hosting provider handles stuff when things go wrong:

- How reliable are their datacenters?

- Do their servers use good hardware? e.g. Hetzner is known to replace HDs on their budget servers with used HDs that have been "retested." Whatever it is, they fail more often than HDs at other hosts.

- Do they have organized procedures I can bet my business on? (DDoS response, hardware failure, power failure, network failure)

- Do they have a dedicated security team?

- Do they have professional, knowledgeable support? Low cost providers are typically lacking, IMO. When shit hits the fan and I need top-notch, professional support at 2am, can I get it?

To me, the only budget host that seems on par with the bigger companies is DigitalOcean.

The reality is that they solve very different things. Part of what you're getting with AWS, GCP, Azure (DO, etc) is API support, and a big set of features. I view the big ones as a PaaS, and 'smaller' players as IaaS. Ultimately IMHO selecting a provider is either a risk mitigation strategy.

OVH in particular is really just phenomenal for dedicated hardware at a very reasonable price. They deploy very quickly (anywhere from 3 minutes to 48 hours, depending), and I've always found them INCREDIBLE to deal with - the last 4 years. Prior to that, when they were the go-to host for lowendwebhosting/cheapwebhost/lowendbox/etc they were a 'best of luck' provider. And that caused grief for many a friend / startup / client.

There's really a risk/reward difference. With AWS and friends, I'm getting incredible scale. With OVH and friends I'm getting tremendous disk space, processing power, but the scaling can be at best a PITA.

Ultimately I'm a big fan of physical to cloud bursting - if that's your need. Monitor your environment from a third party and use that to orchestrate your cloud-bursting strategy if necessary. It's fairly easy to do, if not a tiny bit time consuming.

OVH' stance is "we keep our customer support shitty to keep our prices low".

This is hit-and-miss marketing strategy that works for some, but not for others.

I use them but tbh I miss quality, responsive customer support when things gets bad.

AWS and GCE and the like aren't selling vpses, or dedicated servers, or even cloud services.

They are selling 9s of uptime.

And for the amount of 9s you get, it is the cheapest way to get it.

To add to many other valid points others have provided, OVH is a discount provider and they use second-hand / refurbished hardware at their data centers to keep costs low. I used to manage hundreds of servers on OVH and found that their hardware failed much more frequently than even us-east-1 on EC2. Most common issues were memory and disk related failures. A few times, their techs tripped over a power cord that took a few of our racks down, and they used to have frequent issues with network connectivity (I believe it's gotten better since).

On top of it, you will find that they cut costs on some of the less obvious things. For example, one day we found that their vRacks are not redundant, so a failure in one of them caused hours of downtime of our intranet.

As far as customer support - our account manager was always very helpful and understanding.

> Edit: I would add that we have moved to OVH dedicated servers not VPS.

This invalidates your other points. Your question actually is why are dedicated servers providers not as popular as cloud providers among startups. And I'm sure you'll come with the answer yourself.

> I am aware of the following drawbacks of OVH:

They are all false. OVH is a Public Cloud provider, just like AWS, GCP and Azure.

> * No disk replication i.e unlike EBS or Google disk, if my disk fails, I am at the mercy of my backups (given no RAID)

https://www.ovh.com/world/public-cloud/storage/additional-di...

> * No private network for all the servers by default. Unlike AWS security groups or Google Cloud private networks (which are setup by default), I will have to pay for vRack to setup a private network and is more complicated.

https://www.ovh.com/world/solutions/vrack/

> * Decreased flexibility i.e I may not be able to spin up instances instantaneously or shut down instances to save costs in off peak times.

https://www.ovh.com/world/public-cloud/instances/features/

IAM and the various sdks/aws cli are worth the price alone.

AWS may be the most flexible option out there. And if you need flexibility, that matters. But If you are more established, without changing needs, and can define exactly what yhose needs are, other vendors often will be cheaper.

Startups therefore like AWS - they have too many unknowns. (And too many "don't know what they don't know" factors). More established, mature organizations have more of an option to lose flexibility in exchange for a lower cost.

Personally, the 'free tiers' offered by the big PAAS and cloud providers are good enough for small projects and MVPs-- can't beat free. If and when extra resources are needed, it's usually easier to just start paying at that point rather than to switch platforms entirely.

Free tiers are are extremely effective (albeit crafty) in a "the-first-one-is-on-the-house" kinda way.

AWS, GCE and Azure solve problems that OVH, Hetzner, Online.net etc do not. For many businesses there is a lot of value here. Some of it is because the businesses do not spend that much on server hosting, some because the problems that AWS solves are very valuable.

If you do want to use a bare metal provider like OVH, keep in mind the differences from a service like EC2. You have physical disks. So you either need to implement network storage, or raid, to improve reliability and uptime, or make your software tolerant of data loss. Restoring a bare metal server from a backup has some considerations that a virtual server will not. You also have physical memory, cpus, nics, but in practice these do not fail often. Other areas of the infrastructure are less redundant or reliable because they are a budget provider. You do not have nearly as many physical locations to choose from.

A lot of older businesses are used to buying servers and collocating them instead of renting the whole server and could be saving a lot of money. Or businesses that insist that everything is on prem and never consider the whole cost of power, cooling, security, and connectivity. There are a lot of newer businesses that never really considered something like OVH because they started in AWS, and have only ever considered services like EC2 and competing VPS services, and not bare metal providers like OVH. You also have businesses that know that OVH etc is cheaper, but want to use some new thing for other reasons. It could be to improve their resume, or because this is what investors expect, or because this is what they think the companies that will eventually acquire them will want.

Managing your own servers instead of managed services is hard! It requires admin skills in addition to dev skills on your team, gets seriously hard with increasing levels of security, makes planning and spikes much harder, ... but can also be highly rewarding in terms of costs, mastering your environment, optimization and dev quality, ...

The choice is mostly about how much time and will you have or plan to have on your team for it. You'll have to handle files and database replications, you'll have to update the systems (with reboots), you'll have to fix very tricky issues, ... that would have been handled otherwise.

OVH (and some others like Online.net) have great products with amazing pricings. I have seen a few startups relying on many of their (rather small) servers and it was all great.

One thing to remember is that human time costs much higher than cloud time.