BCDRPulse

Load testingSee how quickly your site answers, before a release.

A load test sends up to six simulated visitors at once at one site you have verified and measures how it responds, so you have a reading before launch day, not after.

100 load tests a month on Pro. 500 on Business.

84.9

requests a second

14 ms

p95 response time

0%

error rate

From the sample report: 1,273 requests, none failed.

The highlights.

p95

and p50 and p99. Percentiles, not just averages.

1 to 6

concurrent users, sent from Cloudflare's network.

30 s

at most for a run. A quick read before a release.

PDF

A branded report with an AI summary.

Yours only

Nothing is tested until its owner has proved it.

A launch, a sale or a mention in the press can bring more visitors in an hour than a normal week. Will the site answer quickly? Will it answer at all? Finding out on the day is the expensive way.

Four steps. From domain to verdict.

  1. Verify the asset

    Every test runs against a domain you've proven you own. No testing systems you don't control.

  2. Send the load

    One to six simulated visitors start together from Cloudflare's edge, and each sends request after request for the 15 to 30 seconds you set.

  3. Measure the response

    Response time, time to first byte, throughput, and error rate are recorded across the whole run.

  4. Read the verdict

    A risk rating set by fixed rules, and an AI summary of what the numbers mean and what to check first.

The numbers, and what they mean.

The report gives p95 and p99 beside the average, because the slow requests are the ones visitors feel. The AI summary points at what to check first.

Load test report

Sample report

Example site

p95 response

14 ms

Throughput

84.9 req/s

Error rate

0%

Consistency

7 / 100

AI summary

Fast, but uneven. A 12 ms average TTFB means the backend responds quickly under light load — yet a p99 of 38 ms against a 327 ms max drags the consistency score to 7/100, the classic signature of occasional resource contention rather than a payload problem. Worth a look: shared-hosting CPU starvation, database query variance, and whether object caching is genuinely active. Is your own site quietly doing the same thing?

A performance assessment, not a penetration test.

Branded, downloadable, shareable: the PDF carries the same page, ready for a review or a vendor questionnaire.

What a load test measures.

Four figures from the sample run, and why each one matters.

14 ms

p95 latency

95% of requests finished faster than this. The honest read on user experience — averages hide the slow outliers that actually frustrate people.

12 ms

TTFB

Time to first byte — how long until the answer starts to arrive, setting up the connection included. Low here means the site is quick off the mark under light load.

84.9 req/s

Throughput

Requests sent per second over the run, failed ones included. The pace at this light load, not a measure of capacity: a cloud load test is the heavier reading.

7 /100

Consistency

How even the response times are. A low score flags scattered latency — caching gaps or resource contention — even when the averages look fine.

What you get.

  • Up to 100 load tests every month on Pro
  • Response time, time to first byte, throughput, and error rate
  • p50, p95, and p99 percentiles — not just averages
  • An AI summary of what the numbers mean and what to check first
  • A downloadable report you can share in a review
  • See all features ›

Edge load test vs. Cloud load test

This is the lighter, edge-run load test. When you need to see how the site holds with 50 to 1,000 users on it, the Business-tier Cloud Load Test runs on dedicated infrastructure.

What is comparedEdge load test (Pro)Cloud load test (Business)
Where it runsCloudflare's edge networkIsolated cloud infrastructure
Concurrency1 to 6 connections50 to 1,000 virtual users
Best forA fast, low-cost sanity checkHolding a level you choose before a launch
EngineLightweight edge workersA load-testing engine on dedicated infrastructure

Is it the right tool?

A strong fit when

  • A launch, a campaign or a sale is coming and you want a number first.
  • You changed hosting, caching or the database and want a before and an after.
  • You want a quick read on response times before each release.

Not the best fit when

  • You need to see the site with hundreds of users on it. Cloud load testing does that.
  • You want to test a site you do not control. Only verified sites can be tested.
  • You need a penetration test: a person trying to break in. We do not offer one.

The Pro load test is deliberately light: up to six concurrent users for up to 30 seconds. It sends real requests, so it runs only against a site you have verified.

Included in Pro and Business.

Pro

100 load tests a month, 25 sites

$49 a month₹4,799 a month in India

Business

500 load tests a month, 100 sites

$199 a month₹19,499 a month in India

In more detail.

What a load test actually measures

A BCDRPulse load test sends a small, controlled group of simulated visitors at one domain you have verified you own, and records four things about how your site answers: response time across percentiles, time to first byte (TTFB), throughput, and error rate. The percentiles matter more than the average. An average of 40 ms can hide a p99 of 400 ms — the slow one-in-a-hundred request that, on a busy day, is a lot of frustrated visitors. That is why the report gives p95 and p99 beside the mean: they are the honest read on what your slowest users feel.

How the edge load test works

The edge load test runs from Cloudflare's global network. The simulated visitors you set, six at most, start together, and each sends one request after another for 15 to 30 seconds while every request is timed; there is no ramp. It is deliberately light and low-cost: a fast sanity check you can run before a release, not a replacement for the Business-tier Cloud Load Test, which holds 50 to 1,000 virtual users from cloud infrastructure. Every run is gated behind verified domain ownership, because driving real traffic at a site you do not control can be unlawful and can take the site down.

When to run one, and what's in the report

Run a load test before a product launch, a marketing push, or any change to hosting, caching, or the database. The deliverable is a downloadable, branded PDF: a risk rating, the full percentile table, throughput and error rate across the whole run, and an AI summary that points to what to check first. Example result: a 14 ms p95 at 84.9 requests per second with zero errors across 1,273 requests — fast, but with a consistency score of 7/100. A low consistency score means response times vary a lot from request to request, which is worth investigating even when the average is fast.

What a load test is not

A load test is a performance assessment, not a penetration test: it measures how your site responds to traffic; it does not probe application logic or attempt exploitation. If your site sits behind a CDN, a fast result may reflect the CDN rather than your own server — keep that in mind when reading the numbers.

Questions people ask first.

Is this a penetration test?

No. Load testing is an automated performance assessment — it measures how your site responds to traffic. It does not probe application logic or attempt exploitation. A penetration test is a separate, human-led engagement. BCDRPulse does not offer one.

How is this different from the Cloud Load Test?

This edge test is capped at low concurrency (6 connections at most) and runs from Cloudflare's network — fast and cheap for a quick read. The Business-tier Cloud Load Test holds 50 to 1,000 virtual users from dedicated infrastructure, to show how your site behaves at a level you choose.

Will it take my site down?

It's a lightweight test, but it does generate real requests, so you run it only against an asset you've verified you own. You set the concurrency and duration.

How many load tests can I run?

Up to 100 every month on the Pro plan, against any of your verified assets. The limit resets at the start of each calendar month.

What is a good p95 response time?

There is no single number — it depends on your site and audience — but for most websites a p95 in the low hundreds of milliseconds or better feels fast. What matters more is that p95 and p99 stay close to your median; a wide gap is a sign of inconsistency, even when the average looks fine.

How is a load test different from uptime monitoring?

Uptime monitoring checks, at intervals, whether your site is up and responding. A load test deliberately sends simulated visitors at it at the same moment and measures how it performs under that load: speed, throughput, and errors.

Can I load test a site behind Cloudflare (or another CDN)?

Yes. Just know that a CDN may serve cached responses, so a fast result can reflect the CDN rather than your own server — keep that in mind when reading the numbers.

Know before launch day.

Signing up is free and starts uptime monitoring for 10 sites. Running load testing needs the Pro plan or Business.

  1. 14 ms p95, 84.9 requests a second and 0% errors are the figures in the sample report, not a customer's site.
  2. Allowances are per calendar month and can be used on any of your verified sites.
  3. Introductory prices, before GST or VAT. Regular prices are $99 and $399 a month (₹9,699 and ₹38,999 in India).