Shield · HTTP flood (layer 7)
When thousands of addresses ask for pages at the same moment, the server can't keep up and customers see an error. Shield recognises the wave and holds it back.
Community is free on one server, with every protection.
On 20 October 2026 an online shop received 18,560 requests in 5 minutes from 17,416 different addresses, almost all with a single request. It was offline for more than an hour.
No address, on its own, did anything unusual: a per-address limit would never have triggered.
A wave of 6,270 addresses in 20 seconds against an online shop protected by Shield. Recognised in about 2 seconds; the site kept answering, with no server errors.
It learns how many new visitors your site usually gets. A wave is measured against your site, not a fixed number.
Visitors already in have a pass and browse as usual. Verified search engines are not stopped.
During a wave, new visitors go through a check in the browser that takes a few seconds. Scripts sending one request each stay out.
Shield does not stop volumetric network DDoS, which fills the server's line before a request reaches the site. That has to be filtered before the server, by the hosting provider or by a network in front of the site (CDN).
With nginx and Apache, if Shield stops the site stays reachable.
It stops HTTP floods (layer 7): more page requests than the server can handle. It does not stop volumetric network DDoS.
Those already in don't. A new visitor arriving during a wave sees a check of a few seconds, then the page.
No. Shield runs on your server, in front of the site.
With Pro and Business, in the panel: start, duration, peak, what Shield refused and why, and which countries and operators it came from.
Shield installs in minutes on your Linux server and starts in observe mode: you see what it would stop before it stops anything.
Install Shield → See the plans →