Applying adds, it never replaces
A preset appends its rules to what is already there. Applying it twice adds a second copy of each rule; there is no button that removes a preset.
Every game server here gets the same filter at the network driver. For the games with a protocol validator, that filter also reads the game's own protocol. For the rest it enforces port rules, packet ceilings and behavioural checks. This page says which is which.
FiveM (GTA RP) preset writes
Minecraft (Java) preset writes
Minecraft (Bedrock) preset writes
Source / CS2 preset writes
TeamSpeak 3 preset writes
Default ports Login, char and map ports
No preset. Add a rule for each port you run.
Rust preset writes
ARK: Survival preset writes
SA-MP / open.mp preset writes
Validators also ship for RDP, DNS and VPN traffic. Ports are defaults; if you moved one, open the one you set. Ceilings are the values each preset ships with and can change without a panel update.
A blocked source ends at the address, a spoofed flood at the transport. The one that needs a validator is the protocol impostor: right port, plausible packet size, but not the protocol the port serves. A rate limit cannot see this attack, because every packet is individually reasonable. The validator can, because the flood never holds the conversation a real client holds. Switch the diagram to a preset-only port to see what changes when there is no validator to read it.
The difference is layer 7, and the protocol abuse that only that gate can see.
| Check | Validated | Preset only |
|---|---|---|
| Layer 3: blocked sources and malformed packets | Yes | Yes |
| Layer 4: spoofed and stateless floods | Yes | Yes |
| Your packet ceilings, per port and per connection | Yes | Yes |
| Signature (zero-day), on UDP rules | Yes | Yes |
| TCP fingerprint, on TCP rules | Yes | Yes |
| Behavioural scoring, timing regularity and reputation | Yes | Yes |
| Layer 7: protocol validation | Yes | No |
| Protocol abuse: invalid payloads, malformed and replayed connects | Yes | No |
| Preset | Rule | Ceiling as shipped | Protocol |
|---|---|---|---|
| FiveM (GTA RP) | UDP 30120 | pps_limit 50000 | Validated |
| FiveM (GTA RP) | TCP 30120 | None | Validated |
| Minecraft (Java) | TCP 25565 | pps_limit 10000, estab_flow_pps 250 | Validated |
| Minecraft (Bedrock) | UDP 19132 | pps_limit 20000 | Validated |
| Rust | UDP 28015 | pps_limit 30000 | Preset only |
| Rust | TCP 28016 | None | Preset only |
| Source / CS2 | UDP 27015 | pps_limit 30000 | Validated |
| SA-MP / open.mp | UDP 7777 | pps_limit 20000 | Preset only |
| ARK: Survival | UDP 7777 | pps_limit 25000 | Preset only |
| ARK: Survival | UDP 27015 | pps_limit 5000 | Preset only |
| TeamSpeak 3 | UDP 9987 | pps_limit 10000 | Validated |
| TeamSpeak 3 | TCP 10011 | None | Validated |
| TeamSpeak 3 | TCP 30033 | None | Validated |
A preset appends its rules to what is already there. Applying it twice adds a second copy of each rule; there is no button that removes a preset.
Close to the cap, a preset can stop partway. The confirmation reports what the preset holds, not what landed, so check the table.
Most preset rules switch Protocol validation on. A rule you add by hand starts with it off, whichever service you pick.
Query ports, RCON, SourceTV and Rust+ are not part of a preset. Add them yourself, and restrict the administrative ones to your own addresses.
A preset is sized for a typical server. A ceiling below your real peak drops players and looks exactly like an attack.
The ceilings above can change without a panel update, so the rules table is the record of what was written.
1
Completing a genuine handshake counts in a source's favour, so does answering a challenge, and so does the irregular timing of a human rather than the metronome of a script.
2
A player who has already proved themselves is not thrown away when the server comes under attack. The filter has a reason to keep them that a rate limit does not.
3
An unknown source arriving during an attack has to earn its way in at the moment the bar is highest. That is also why pausing protection mid-attack is worse than it sounds.
| Game | Start on | Why |
|---|---|---|
| FiveM | MicroCore VDS, up to 32 players | Room for a framework such as ESX or QBCore with its MySQL or MariaDB database on the same machine. Its cores are never shared, so the main game thread keeps its tick while the database and streaming run beside it. |
| Minecraft | Mini VPS, a small world | A small world and a few plugins, with memory left over for the operating system and for what the Java runtime uses outside the heap. Giving the heap every gigabyte on the machine is the common mistake. |
| CS2 and Source | Pro VPS, one busy server | A Source server is one process. The cores around it carry SourceTV, RCON tools and updates without crowding the game. |
| TeamSpeak | Nano VPS, on its own | A single binary with an SQLite database beside it, undemanding next to a game server, so the smallest plan carries it. |
| Rust | Pro VPS, one busy server | Memory is what Rust runs out of first, and it climbs through each wipe as players build. Start with headroom and check the RustDedicated process at the end of a cycle. |
| ARK | Pro VPS, one map | A bare map needs several gigabytes before anyone connects, and a busy modded map considerably more. Start with room to grow and measure. |
| SA-MP and open.mp | Nano VPS, to start | Start on Nano and resize as the server grows. CPU and traffic changes apply straight away; memory and disk at the next restart. |
Each game page explains the sizing in full.
Validators ship for FiveM, Minecraft, Source, rAthena, TeamSpeak, RDP, DNS and VPN. Source covers CS2, CS:GO, TF2, Garry's Mod, and Minecraft covers Java and Bedrock.
Yes, by port presets, packet ceilings, the layer 3 and layer 4 filter and the behavioural engines. They are not protocol validated: no validator ships for them, and we do not claim one.
No. Every plan in every region sits behind 17 Tbps of upstream capacity and is filtered by Prism. There is nothing to add at checkout.
Yes. Add a rule for each port it uses. The layer 3 and layer 4 checks, your ceilings and the behavioural engines apply to every rule. If the service has no validator, the Protocol validation switch renders dim and has no effect.
It starts off on every rule you create by hand, whichever service you pick, and most preset rules switch it on. A validator drops anything that does not speak the protocol, so it is left for you to turn on while you can watch the result.
It can, if the client speaks the protocol loosely. Address lookup on the Attacks tab names the check that dropped a player, and you can switch validation off for that one rule.
Every plan carries the same filter and the same 17 Tbps upstream, enabled from the first boot.