IT Capacity Planning: Throughput, Latency & Scaling

What you'll build

What you will be able to do

Each lab ends with a concrete result that you build yourself and the checker confirms, cell by cell.

• Use L = λ × W to turn the peak load and latency of five Corvo Tickets services into requests in flight, and check whether their worker slots are enough.
• Calculate the utilisation and mean response time of Corvo Tickets' single database writer for seven load scenarios and find the highest load that still meets the latency target.
• Take 40 response times of the Corvo Tickets checkout endpoint, compute p50, p95 and p99 with PERCENTILE.INC, check the SLO and draw the latency histogram.
• Work out how long five Corvo Tickets data transfers take, whether they fit the 4-hour night window, and which link speed each one would need.
• Measure the compound monthly growth of Corvo Tickets' object storage, forecast the next 12 months and work out when you must order new capacity.
• Size five Corvo Tickets services for the peak with 30% headroom and one spare, then turn the instances into a monthly bill and a cost per 1,000 requests.
• Combine six tiers of the Corvo Tickets checkout path – some redundant, some not – into one system availability, downtime per year and number of nines.
• Analyse a 10-step load test of the Corvo Tickets checkout, find where throughput stops growing, pick the highest load that still meets the p95 SLO and chart the knee.