Maybe. We aren’t out to provide the cheapest possible service. In fact, we have absolutely no interest in doing so. Our goal to is to provide a high-quality service at a fair price, and make it scale so we can put that service in the hands of as many people as possible, many of whom couldn't otherwise afford it.
With that said, the most common reason people feel that we're more expensive than an alternative is that they are comparing us to a different type of service.
We provide a fully managed environment, including frequent security updates, careful performance tuning, and 24x7 response to crashes and outages for everything underneath your code. The tools you need to build your site are already there, ready to be used and customized however you want. We have dedicated infrastructure to build a constant stream of updates, more infrastructure that runs regression tests on those updates to make sure things are working before they get to you, then we let you pick just how aggressively you want those updates applied.
With an unmanaged VPS, the provider maintains the hardware (hopefully), but doesn’t care if your VPS is up or down. The full responsibility of installing and maintaining the OS, the system configuration, all the packages and utilities, and the entire application stack falls on you. Every critical security update. Every urgent kernel patch. Every weird error in the system log. Every crash in the middle of the night. Every traffic surge. You have to know about it, you have to know what to do about it, and you have to do it. Often on a deadline. And that responsibility never goes away. It never even takes a day off.
Occasionally, people comparing us to a VPS zero-value that, because it hasn't occurred to them, or because they plan to do it themselves (some of it, anyway — system administration is a well-paying full-time job) and haven’t placed a dollar value on their time. But VPS providers sure don't zero-value it. The price difference between an unmanaged VPS and a fully-managed VPS from the same provider is often $50-200/month. Already, our pricing looks a lot better!
Second, the lowest VPS deals come from brand new providers who lose money on every sale. They’re buying market share by selling below their cost. We don’t compete with them on price. We just wait for them to go out of business.
Third, VPS providers use all sorts of tricks to make it seem like they’re providing a lot more than they really are. Hopefully you’re already convinced, but if you’re in the mood to keep reading, the rest of this FAQ entry contains a partial (but long) list of such tricks and what we do differently.
VPSs and RAM:
- A fair bit of the RAM in a VPS goes to the kernel. Behind the scenes, providers use a technique called memory deduplication to sell the same memory to 100+ VPS customers on a single machine. We don't charge you for kernel memory at all.
- If you want your application to perform well, you'll need a fair amount of memory for disk cache. For a nontrivial application, usually 25% or so of the total memory, although the system will always try to use as much as it can. With a VPS, that comes out of the RAM you're paying for. At NearlyFreeSpeech.NET, it does not.
- The memory a VPS provides you is a limit. You can't exceed it. If your VPS runs out of RAM, when you get a surge of activity, it won’t be able to scale and may as well be down. That means you have to pay for the most memory you'll ever need to handle the highest load you anticipate ever having. The other 99% of the time, you're still paying for all that RAM, but it's sitting there unused. We charge you for the memory you’re using right now. If you need more, it’ll be there, if you need less, you don’t pay for what you’re not using.
VPSs and storage space
- Modern filesystems require at least 15-30% slack space to perform well, so that comes right off the top of a VPSs storage allocation. We only charge for the space you're actually, currently using. If you delete files, your bill goes down.
- Then you install the whole OS into it, eating up even more. (And in the background, they quietly deduplicate that too, so they can sell the same disk space over and over.) We don't charge you for the space used by the OS, the system tools, utilities, and development tools (or the entire application stack, if you use one of the ones we provide, like PHP).
- You have to leave room for growth. It's not unusual to see a VPS with only 10-15% of its allocated disk space actually used by application data. Here, the space will be there if you need it, but there’s no need to pay for it until you do, or after you’re done with it.
- VPS storage usually isn't backed up by default. That costs extra and is often a "last backup only" approach. Our storage pricing for sites includes extensive, frequent snapshots and backups — onsite and offsite — at no additional charge. (If you need a restore performed, that takes a human, so there is usually a small charge for that at that time.)
- At most providers, storage is directly connected to the machine hosting a VPS. That's good for performance, but not so good for reliability. If the machine hosting a VPS dies, well, see above about backups. We use redundant network storage built entirely from SSDs with no single point of failure for all member sites and MySQL data.
VPS overselling
- VPSs often measure CPU based on “virtual” CPU cores (“vCPUs”). As with RAM, there are a lot more “vCPUs” in VPSs than there are actual CPUs in the machine running them. If too many people try to use them at once, things slow way down, and there’s usually no way for you to figure out why. With us, most sites don’t pay for CPU at all, but those that do pay only for what they’re actually using at that moment.
- VPS providers rarely support live migration. If you’re on a heavily-loaded server (or the server crashes) at a cheap VPS provider, tough luck. Decent providers will tell you a server is overloaded and offer to move you, but without live migration, that means a long downtime while they copy your disk images around. NearlyFreeSpeech.NET uses dynamic load balancing directly informed in real time by the same resource metrics used for billing. If a server gets overloaded here, load automatically shifts. If it’s still (or very) overloaded, somebody gets paged to check it out or allocate extra hardware for it if needed. If a server goes down, affected sites automatically migrate to other servers within about 60 seconds.
- Check the VPS provider’s Terms of Service for words like “excessive resource usage.” If you find them, it’s often code for “if you actually try to use what you paid for for any length of time, we will promptly put a stop to that.” Here, charging only for the resources we actually provide creates a financial incentive for overprovisioning instead of overselling.
We are not trying to say all VPSs are bad, or even that they’re all a bad deal. There are some applications where a VPS is a better fit than our service. And if you don’t need or want all that we provide, and don’t mind doing a lot of work yourself, a low-end VPS may save you some money. But in an awful lot of cases, for sites big and small, we provide an excellent value that no VPS can match.