Why my Minecraft hosting comparison site has no scores
Most hosting comparison sites give each provider a score. "Host A: 9.4/10." I run MCHostingHub, a comparison of Minecraft server hosts, and it does not do that. This post is about why, and about the technical side of Min
Most hosting comparison sites give each provider a score. "Host A: 9.4/10." I run MCHostingHub, a comparison of Minecraft server hosts, and it does not do that. This post is about why, and about the technical side of Minecraft hosting that makes comparing plans harder than it looks.
A Minecraft server has 50 milliseconds
A Minecraft server runs the world in ticks, 20 per second. Each tick has 50 ms to update every loaded chunk, entity, redstone circuit and player. If a tick takes longer, the server falls behind, TPS drops below 20, and players see blocks reappear and mobs teleport.
1000 ms / 20 ticks = 50 ms per tick
tick takes 70 ms -> about 14 TPS, players notice
Most of that tick runs on one main thread. Paper and other forks move some work, like chunk loading, onto other threads, but the core game loop is still single-threaded. So the thing that matters most is how fast one CPU core is.
That has a direct effect on hosting. A host can put a 4 GB server on a slow, old CPU shared with dozens of other servers, and a 4 GB plan on a fast modern CPU will run much better. Both look the same on a plan page if you only read the RAM number.
RAM is how hosts sell, not how servers perform
Almost every Minecraft host sells plans by gigabytes of RAM. It is easy to understand and easy to compare. It is also only loosely related to how the server will run.
More RAM helps up to a point. A modpack with 300 mods needs a lot. A vanilla survival server with ten friends does not. Past what the server needs, extra RAM does nothing, and a very large heap can make garbage collection pauses longer if the JVM flags are not tuned. This is why the Aikar's flags that most hosts use exist in the first place.
There is a separate guide on how much RAM a Minecraft server needs, because "buy more RAM" is the most common way people overpay.
Still, RAM is what everyone sells by, so price per GB is a useful way to sort plans as a first step:
const plans = [
{ host: "A", ram: 4, monthly: 10.00, renews: 10.00 },
{ host: "B", ram: 6, monthly: 9.00, renews: 15.00 },
];
for (const p of plans) {
const first = (p.monthly / p.ram).toFixed(2);
const after = (p.renews / p.ram).toFixed(2);
console.log(`${p.host}: $${first}/GB first month, $${after}/GB after`);
}
// A: $2.50/GB first month, $2.50/GB after
// B: $1.50/GB first month, $2.50/GB after
The renewal column matters. A plan that is cheap for the first month can be the same price as everyone else from month two. Comparing first-month prices alone makes some hosts look much better than they are.
Why not just score them?
To give a host 9.2 out of 10, you need to turn CPU, price, support, locations and reviews into numbers and weight them. Every weight is a guess. Is support worth 15% or 25%? Is a cheaper plan on a slower CPU better or worse than the reverse? The final number looks precise, but it mostly reflects the weights someone picked.
There is also a hardware problem. Hosts often do not say which CPU a given node uses, and the same plan can land on different machines. I have not benchmarked every provider's hardware, so giving them a "performance score" would be making it up.
So the site groups hosts by what they fit instead: cheap Minecraft server hosting for small budgets, hosts for SMPs, modpacks, plugins, proxy networks and Bedrock crossplay. Each page says why a host is there, and you can decide what matters to you.
Data that can be checked
The parts that can be checked, get checked:
- Prices come from each provider's own pricing page and are checked every week. Anything that cannot be confirmed automatically is reviewed by hand.
- Bedrock support is checked against GeyserMC's list of supported hosting providers. Each host is marked as built into the panel, working with manual setup, or not on the list.
- Things that cannot be measured, like how fast support answers or whether nodes are oversold, come from what people say in places like r/admincraft. Those are treated as signals, not facts, and the page says so.
The full method is written down on the how we rank page. One more rule: affiliate commission does not change the order. Some links on the site are affiliate links, and the site has a disclosure page that explains how.
What I would tell anyone building a comparison site
- Show the data you have and say where it came from. Users can handle "we have not tested this".
- Normalize prices so plans can be compared, and include the renewal price.
- Do not invent precision. A grouped list with reasons is more honest than a ranked list with decimals.
- Write the method down publicly. It keeps you consistent, and it is the page people check when they wonder if you are just selling something.
If you are picking a host for your own server, the full Minecraft hosting comparison has every provider in one table with prices, CPUs and Bedrock support. Feedback on what is missing is welcome in the comments.
Originally published by Dev.to WebDev. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.