Proven effectiveness

The same attack. Without PAKKT and with.

A multi-source flood of 3.95 billion packets hits the same server. On the left, nothing filters it. On the right, XDP + nftables + Auto Shield drop it. Watch where the attack dies.

WITHOUT PAKKT The server takes everything
WITH PAKKT The attack dies before the server
Metric
Without PAKKT
With PAKKT
Packets reaching the server
3.91 billion
42.6 million
Share of attack filtered
0%
98.9%
Server state
Saturated · packet loss
Healthy · normal latency
Player latency
Timeouts
< 1 ms added

Illustrative scenario based on a typical multi-source volumetric flood

Why it's honest

Same attack, same machine, only one setting changes.

The agent measures incoming traffic in both cases. The difference is in the action: without filtering, every attack packet reaches the kernel and burns CPU; with PAKKT, XDP drops it inside the network driver, before any memory allocation.

That is the whole point of filtering as early as possible: whatever never reaches your server costs it nothing. The rest, the legitimate traffic, passes through with added latency under a millisecond.

Keep going

Explore the platform

Real-time monitoring Live dashboard, GeoIP world map, top IPs and ports. Discover → XDP + nftables rules Two independent kernel layers, one panel. Discover → Auto Shield Automatic attack detection and IP banning. Discover → Geo-blocking Allow or block traffic by country, via CIDR sets. Discover →
FAQ

Frequently asked questions

Straight answers, no marketing fog.

Is this comparison realistic? expand_more
Yes. The PAKKT agent measures incoming traffic in both cases. It is the same attack on the same machine. The only difference is what gets dropped: in observation mode nothing is filtered and everything reaches the server; with the shield active, XDP and nftables drop the attack before the kernel. The orders of magnitude (millions of legitimate packets, billions of attack packets) match a typical volumetric flood against a game server.
Why talk about "packets reaching the server" rather than bandwidth? expand_more
Because a volumetric attack saturates packet-processing capacity (pps) long before raw bandwidth. XDP drops packets inside the network driver, before the kernel allocates any memory: whatever never reaches the stack costs almost nothing in CPU. The right effectiveness metric is therefore the number of packets that never reach your server at all.
What does "well tuned" mean? Does protection work right after install? expand_more
Baseline protections apply right from installation (nftables system rules, anti-amplification, anti-self-block guard). Maximum effectiveness comes from setting the right XDP/nftables rules for your services and arming the Auto Shield with a trust zone that fits. The panel and the template marketplace make this tuning fast, and monitoring shows you the effect live.
Does PAKKT guarantee that an attack will never get through? expand_more
No serious protection promises absolute invulnerability. PAKKT moves the blocking point as early as possible (XDP, before the kernel) and automates mitigation, which absorbs the vast majority of common volumetric attacks with sub-millisecond added latency. For extreme volumes that saturate the network link itself, PAKKT combines with your host's network-level protection.

Put your server on the right side.

From 3 € / server / month. One-command install, protection active in minutes. Free 7-day trial, no credit card required.

Protect my server See the dashboard in detail