Index
A load test needs to represent the expected behavior of visitors, not just produce lots of requests. Hiring must relate volume, journeys and approval criteria to the planned campaign.
How to evaluate this decision
Map pages, searches, forms and external dependencies. Differentiate simultaneous visitors from arrival rate and consider session duration. Test in an authorized environment, with limits to not affect third parties. A cached quick page does not prove that the quote submission or stock query support the same volume.
Criteria for comparing proposals
- Model: distribute traffic between journeys and representative data.
- Limits: define acceptable times, error rate and recovery capacity.
- Diagnosis: collect evidence to locate the bottleneck in addition to the aggregate result.
A scenario to discuss with the supplier
Hypothetical example: the campaign takes users to a dynamic search. Testing only the static homepage leaves the main path unevaluated.
What to validate upon delivery
Execute growth, peak, and recovery as per the roadmap. Check data integrity and behavior after load, not just availability during testing.
Prepare the conversation about the project
Provide Quantum9 with an estimate of traffic, journeys and infrastructure. The result should state the conditions tested and limits observed, without promising unlimited capacity.
Software engineering and quality · Map the company's priority