Peak Trading Load Testing Agent
Load tests the store for sale events and finds bottlenecks.
- Industry
- Retail
- Department
- Engineering
- Team
- Platform and Reliability
- Kind
- Investigator
- Approver
- Head of Platform
- Code
- ENG-19
It runs load tests on the online store, checkout, availability and order capture ahead of sale events and launches, at last year's peak plus headroom, and explains where and why the system bends, such as the promotion engine, the stock check or the payment callback. It recommends the fixes and the virtual queue settings for the event. **The problem it solves:** Where and why the store's systems slow down or fail under sale-event load.
How it works
6 steps, and a person signs off before step 6.
- 01Looks it up
Finds the record the request is about in the monitoring service: last year's peak metrics for the event and the traces and metrics from the load test runs..
- 02Reads the request
Works out what is being asked and what a complete answer needs.
- 03Checks the facts
Checks figures, dates and rules against the source records, not from memory.
- 04Works it through
Weighs competing explanations against the evidence and shows which one holds up.
- 05Asks a person
Hands the work to Head of Platform with its reasoning before anything goes out.
- 06Acts
Once approved, it acts in the issue tracker: it files the bottleneck fixes for the event and records the test results. and records what was done.
What it works with
Your knowledge, your systems, your people.
Knows
The performance guide: last year's peak traffic and orders per minute by event, the load test scenarios (browse, search, add to basket, checkout), the service capacity limits, the virtual queue and the test environment rules.
Reads from
the monitoring service: last year's peak metrics for the event and the traces and metrics from the load test runs.
Acts in
the issue tracker: it files the bottleneck fixes for the event and records the test results.
Hands back
An object with target_load, results (a list of scenario, load, p95_ms, error_rate and passed), bottleneck (where, likely_cause and evidence), alternative_causes (a list), fixes (a list) and queue_recommendation.
Asks a person when
The test must run against production, or the event cannot reach its target and the queue trigger must be lowered.
Settings
Your limits, set by your managers.
| Setting | Default | What it controls |
|---|---|---|
| Test above last year's peak by (%) | 50 | The target load is last year's peak plus this. |
| Checkout p95 latency target (ms) | 1500 | Checkout must answer within this at the target load. |
| Highest error rate at target load (%) | 0.5 | A higher error rate fails the test. |
| Turn on the virtual queue at (orders a minute) | 4000 | The virtual queue opens at this order rate. |
Limits are settings, not code: a manager changes them without a developer.
Rules it always follows
- Test at last year's peak plus 50 percent; checkout must stay under 1500 ms at p95 with errors below 0.5 percent.
- Find where the system bends first and weigh at least two explanations, such as database locks versus a downstream timeout.
- Run load tests only in the performance environment or with the payment provider's test mode; never place real orders.
- Recommend the virtual queue trigger and say what the test shows about 4000.
- Rerun the failing scenario after each fix and report the change.
See the Peak Trading Load Testing Agent agent on your own work.
We’ll run it on a sample of your work and show you what your approver would see.