April 23rd, 2020 at 10:12 PM
We host on Ramnode. They are what's known as a low end box VPS provider, meaning that they cater to the budget end of the market. That being said, they are one of the best Low End Box providers in existence. They reliability is very good. Their prices aren't the cheapest that you'll find, but you get what you pay for, and Ramnode provides a lot of value for the money.
However, a quick look at their plans and you'll realize that there are plans that give you 4 cores for $10/month. How?
Even better, their premium VPS line (much faster CPUs) offers 4 cores for $48/month. This is actually a fantastic price. You won't find something this cheap at many VPS providers. Some deeper digging leads to a few questions, however.
Ramnode's premium VPS line uses blazing fast 3.7ghz CPUs (and that's the base clock!) However, they achieved this primarily with Intel E3's (and occasionally generic E-series) CPUs. E3's are strictly single socket CPUs with a maximum of 4 cores and 8 threads. This 4-core VPS is, quite literally, being hosted on a server that only has four cores for ALL VPSs available. They are, in theory, lending you the entire server's CPU for $48 dollars, or are they really?
Not quite. Firstly, for all we know, they could be using the E-series generics that have 6 cores and 12 threads (not unlikely). But the real kicker here is that these "cores" they are giving you are, in actuality, just threads. You are getting about 50% of the threads on the server, but not 100% of the cores, per se.
Even this is a little complicated. You're sharing these threads with multiple other servers on the hardware. They put a number of them on there. These aren't dedicated cores. If your neighboring nodes aren't behaving, your performance will suffer. Some VPS providers are much worse about this than others. Ramnode hasn't given us many problems.
Additionally, hyperthreading usually results in about a 15-30% increase in performance. In a gross simplification, if your VPS is the only VPS using significant CPU, you will get 100% of that core. When another VPS gets put on the other thread of an individual CPU, hyperthreading kicks in and your VPS only gets about 55-65% of its ordinary level of performance from that core (since it's being split between two VPSs).
In other words, "4" cores means anywhere between four full cores, and merely approximately ~2 cores of performance if the server is busy.
So then, what is their business model?
Their Acceptable Use Policy offers some clues.
In other words, your VPS is strictly fair-share CPU. Ramnode can do this because the vast majority of web servers very rarely use a large amount of CPU consistently. They need to burst quickly. During peak hours, you'll need it. But in general, you won't be using anywhere near the 100% of your CPU allowance. Because of this, they can offer inexpensive burst performance with a decent amount of RAM, and they can do so on blazing-fast E3's with extremely high clock speeds. You get fantastic performance when you need it, but at the prices of a lesser VPS because you likely won't be using much resources most of the time.
What you don't get is the ability to treat it like a dedicated server. In essense, it's shared hosting on steroids. You can't max these things out constantly.
What if you need more?
That's when it gets complicated. And this is where no low end box provider will meet your needs. You will pay 2-4X as much to get the same number of CPU cores on a dedicated VPS. On these VPSs, these cores are solely assigned to you. You are allowed to max them out 100% of the time, and this will not impact surrounding nodes (so long as you aren't thrashing the IO nonstop).
I'll add that a CPU core can usually handle a surprisingly large amount of traffic. You rarely need as much as you think that you do for standard websites.
But with that said, dedicated VPSs are expensive. You will wind up paying much more because the servers you are hosted on are rarely utilized anywhere near their maximum CPU capacity. This is an advantage to you for performance and reliability, but is a disadvantage for your wallet. The host cannot put nearly as many VPSs on the same server, and thus must charge more.
In essense, you want to assess your needs and find the right VPS for your application and traffic level. If you don't plan on using CPU heavily, you can get a great deal on a low end box provider (such as Ramnode). They make good use of their server resources and are able to offer great prices. Otherwise, get a dedicated VPS to avoid risking suspension. Not all VPSs are made the same.
However, a quick look at their plans and you'll realize that there are plans that give you 4 cores for $10/month. How?
Even better, their premium VPS line (much faster CPUs) offers 4 cores for $48/month. This is actually a fantastic price. You won't find something this cheap at many VPS providers. Some deeper digging leads to a few questions, however.
Ramnode's premium VPS line uses blazing fast 3.7ghz CPUs (and that's the base clock!) However, they achieved this primarily with Intel E3's (and occasionally generic E-series) CPUs. E3's are strictly single socket CPUs with a maximum of 4 cores and 8 threads. This 4-core VPS is, quite literally, being hosted on a server that only has four cores for ALL VPSs available. They are, in theory, lending you the entire server's CPU for $48 dollars, or are they really?
Not quite. Firstly, for all we know, they could be using the E-series generics that have 6 cores and 12 threads (not unlikely). But the real kicker here is that these "cores" they are giving you are, in actuality, just threads. You are getting about 50% of the threads on the server, but not 100% of the cores, per se.
Even this is a little complicated. You're sharing these threads with multiple other servers on the hardware. They put a number of them on there. These aren't dedicated cores. If your neighboring nodes aren't behaving, your performance will suffer. Some VPS providers are much worse about this than others. Ramnode hasn't given us many problems.
Additionally, hyperthreading usually results in about a 15-30% increase in performance. In a gross simplification, if your VPS is the only VPS using significant CPU, you will get 100% of that core. When another VPS gets put on the other thread of an individual CPU, hyperthreading kicks in and your VPS only gets about 55-65% of its ordinary level of performance from that core (since it's being split between two VPSs).
In other words, "4" cores means anywhere between four full cores, and merely approximately ~2 cores of performance if the server is busy.
So then, what is their business model?
Their Acceptable Use Policy offers some clues.
Quote:As in all virtual server environments, CPU cores and disk I/O are shared among multiple VPSs. If your VPS is consistently maxing out one full core or more and impacting the performance of other client servers sharing the same host node, we may ask you to reduce your usage. Load may be generated by CPU usage and/or I/O. Disruptive load may result in a reboot, shutdown, and/or suspension of the VPS regardless of the time lapse involved.
In other words, your VPS is strictly fair-share CPU. Ramnode can do this because the vast majority of web servers very rarely use a large amount of CPU consistently. They need to burst quickly. During peak hours, you'll need it. But in general, you won't be using anywhere near the 100% of your CPU allowance. Because of this, they can offer inexpensive burst performance with a decent amount of RAM, and they can do so on blazing-fast E3's with extremely high clock speeds. You get fantastic performance when you need it, but at the prices of a lesser VPS because you likely won't be using much resources most of the time.
What you don't get is the ability to treat it like a dedicated server. In essense, it's shared hosting on steroids. You can't max these things out constantly.
What if you need more?
That's when it gets complicated. And this is where no low end box provider will meet your needs. You will pay 2-4X as much to get the same number of CPU cores on a dedicated VPS. On these VPSs, these cores are solely assigned to you. You are allowed to max them out 100% of the time, and this will not impact surrounding nodes (so long as you aren't thrashing the IO nonstop).
I'll add that a CPU core can usually handle a surprisingly large amount of traffic. You rarely need as much as you think that you do for standard websites.
But with that said, dedicated VPSs are expensive. You will wind up paying much more because the servers you are hosted on are rarely utilized anywhere near their maximum CPU capacity. This is an advantage to you for performance and reliability, but is a disadvantage for your wallet. The host cannot put nearly as many VPSs on the same server, and thus must charge more.
In essense, you want to assess your needs and find the right VPS for your application and traffic level. If you don't plan on using CPU heavily, you can get a great deal on a low end box provider (such as Ramnode). They make good use of their server resources and are able to offer great prices. Otherwise, get a dedicated VPS to avoid risking suspension. Not all VPSs are made the same.
![[-]](https://makestation.net/images/core//collapse.png)
