Highload: when the system has outgrown its clothes

Tick what you recognise

Five statements — and an honest verdict: do you need this service and which package to start with.

Load graph — move the RPS

This is how we find the system’s limit before Black Friday. The slider is requests per second; the graph shows response time and the point where caches and queues are needed.

Not “a service” but concrete artefacts

What you hold at the end — we remove the fear of “paid and got nothing”.

Pick a package — the scope highlights

Greyed out — what is not in the package. Ranges are medians of our projects.

• Load testing — traffic profile, finding the limits
• Architecture — queues, caches, async, sharding
• Database — indexes, replication, partitioning
• Fault tolerance — redundancy, graceful degradation
• Observability — bottlenecks visible before an incident
• Drills — simulating peaks and failures before the season

Move the sliders — the range updates instantly

A benchmark based on medians of our projects. The exact estimate comes after the solution map.

Switch the scenario — we compare the alternatives

Sometimes the better deal is not ours — and we say so first.

What we do and what you get every week

Hover a point. Only artefacts, no “work in progress”.

What usually goes together