Fictional practice, not an employer interview question.
The following is a fictional capacity model. An origin server can handle 250 requests per second under a chosen test workload. Every request either comes from the cache or reaches the origin; assume no retries and no request combining. This deliberately simplified model lets you check the arithmetic before adding realistic complications.
| Condition | Incoming requests per second | Served by cache | Reaching the origin |
|---|---|---|---|
| Normal | 1,000 | 90% = 900 | 100 |
| Degraded cache | 1,000 | 60% = 600 | 400 |
Work through the case
Normal origin traffic is 100 requests per second, below the 250-request capacity. In the degraded case it rises to 400, exceeding capacity by 150 requests per second. Origin traffic has quadrupled even though incoming traffic has not changed.
At 1,000 incoming requests per second, at least 75% must be served before reaching this origin to remain at its stated limit: 1,000 × (1 − 0.75) = 250. Alternatively, with a 60% cache-hit rate, the modeled limit is 250 ÷ 0.40 = 625 incoming requests per second. These are arithmetic boundaries under the assumptions, not safe production operating limits.
Prepare your application
Prepare one example that behaves differently when something goes wrong. A networking or infrastructure application can show a safe local load test. A product or customer-facing application can explain the user consequence and the decision to stop or expand a change. The useful evidence is your reasoning, not a claim of internet-scale experience.
Start with a role and a problem you can test
Use Cloudflare’s early-talent route to find the actual internship. Read the particular team, term, location and required skills; a link to a student program is not an application for every engineering function. For the example here, no Cloudflare account, paid service or live target is needed.
Choose a fallback without ignoring correctness
You could reduce accepted traffic, increase tested origin capacity or serve eligible stale content. Those choices have different user consequences. In an invented catalog, an older product description may be tolerable while an outdated stock-availability answer is not. Define which response must be fresh rather than applying one cache rule to every request.
Build the test against a service you own and bound the request rate. First measure what your toy origin can handle for the chosen response and hardware. Then vary the cache-hit rate and include a cold-cache run. Record origin requests, failed requests and response time together. A lower latency among successful responses can hide the requests that failed outright.
Make the project credible in an application
A student-sized description is: “Built a local cache/origin test and compared normal, degraded and cold-cache runs; documented the point where requests exceeded the measured origin limit.” Supply the workload, hardware and results you actually used. Do not replace this with an unsupported percentage improvement or claim of handling a regional outage.
Your short write-up should recommend a bounded next test and a condition that would make you stop it. For a customer-facing role, translate the same findings into what customers would notice and what information they need. Ask the interviewer how an intern’s change is tested before it reaches real traffic.
Sources and related guides
Official sources
Official sources and current instructions
The source collection was revised September 5, 2026. These links help you find the employer’s current guidance. This guide does not establish a current opening, eligibility, deadline, or interview process.
About this guide
By InternshipDeadlines · AI-assisted editorial adaptation · Updated September 17, 2026
Original practice material, not a reported employer interview question. Scenarios are hypothetical. Original fictional capacity case; not a Cloudflare benchmark or assessment.
Published by InternshipDeadlines. Guide updated . How we use sources.
Cite this guide
InternshipDeadlines. Cloudflare internship application: explain performance under failure. Updated . https://internshipdeadlines.com/career-guides/cloudflare/
Use the dates attached to the sources behind your claim. An editorial update does not reverify every source or confirm that an internship is open.