Find out how the system copes at peak, beforehand
We simulate your peak traffic, such as Black Friday or the start of term, find what gives way first and show what to adjust.
Who it's for
For companies with a peak date in the diary and no idea how the system will behave.
What it solves
- At the last peak the system slowed down, and nobody knows which part gave way.
- The campaign is scheduled, but nobody has tested the system at that volume.
- The answer to slowness has always been a bigger server, and the cloud bill keeps growing.
- A test run once became a number nobody can reproduce.
What we do
- Load model
- How many users, doing what and at what rate, based on your data and the business forecast.
- Peak scenarios
- Gradual ramp-up, sustained peak and short bursts, over the flows that matter most on the day.
- Test environment
- An environment close to production; testing in production only with an agreed window and approval.
- Third-party services
- Bank, PSP and government APIs replaced by simulations, so nobody takes load they have not agreed to.
- Finding bottlenecks
- Database, queues, external calls and code measured during the test to find what gives way first.
- Repeatability
- Scripts versioned in the repository, to repeat the test after each change and before each peak.
What you get
- A documented load model, with the assumptions behind each figure
- Test scripts versioned in your repository
- A bottleneck report in order of impact, with evidence
- A list of recommended changes, split into code and infrastructure
- A before-and-after comparison of the changes, on the same scenario
How we do it
Assumptions
The expected peak, the critical flows and what slow means for your business.
Preparation
Environment, test data, third-party simulations and measurement set up.
Running
Scenarios run in increasing order, with the team watching the dashboards.
Analysis
Each bottleneck tied to evidence and a proposed change.
Second round
The same scenario after the changes, to measure the difference.
Technology examples
- k6
- Gatling
- Apache JMeter
- Locust
- Grafana
- OpenTelemetry
Related services
FAQ
Will you tell us how many users the system can take?
We tell you how it behaved in the tested scenario and what gave way first. That result holds for that scenario and that environment; it is not a promise for every situation.
Can you test directly in production?
Sometimes, within an agreed window and with care around third-party services. In most cases, an environment close to production is enough and carries less risk.
Who decides the peak volume?
Your company, based on its history and the business forecast. We turn that forecast into a load model and show the assumptions, so everyone knows what was tested.
What if the problem is in the database?
Then queries, indexes and database configuration go on the list of changes, and the work can continue under databases.
Write to Balkan
Talk to us about your project