How long before a launch should I test?

Two to three weeks before, then again after the fixes. A test the night before launch is a coin flip with a report attached.

Which pages should I test?

The ones that carry revenue and traffic: landing, listing or search, product, and the checkout API. A fully cached homepage passes any test and teaches you nothing.

Does the test include the CDN?

Only if requests reach the origin. With a cache in front, the run measures the edge. Cache-busting forces origin traffic, which is the right question for a failover drill.

Website stresser for the pages your users open

Send requests at real URLs and see which page degrades first, with latency percentiles and error classes in the same report.

Start a free test See plans

Updated November 8, 2026, 14:54 (GMT+3) · reviewed for accuracy

Website stresser for the pages your users open: Test pages, not just a host

A website stresser takes a URL instead of an IP, so the test measures what a user waits for: request handling, cache behaviour and upstream latency. Start with the page that has the least cache, then compare it with a Layer 7 stresser run.

Start with the page that has the least cache in front of it. The homepage usually survives; the checkout usually does not. Pair the result with a Layer 7 stresser run for detail.

Website stresser for the pages your users open: Cached and uncached pages

The website stresser report is most useful as a comparison between pages, not between runs. A cached page and an uncached one usually degrade at different rates, and the gap is the value the CDN provides. Run both through the free stresser before changing anything.

Behind Cloudflare or another CDN, add cache-bypass requests so the test reaches the origin rather than the edge.

Website stresser for the pages your users open: What to record after the run

Record the URL, rate, delivery ratio and first error class after each website stresser run. The free stresser trial is enough to get the first data point without a card.

Before your first stresser run

Own the target. Verify it through DNS or the panel before any traffic moves.
Pick one question. Does the endpoint fail, and at what rate? One method per run answers it.
Keep the report. The run lives in the panel so a later test can confirm the change.