IT Capacity Planning: Throughput, Latency & Scaling
Little's Law, M/M/1 queueing delay, p95/p99 latency, transfer times, storage growth, servers for the peak, serial/parallel availability, the load-test knee and cost per 1,000 requests: 8 labs on one ticketing platform.
The first lesson is free — no account needed.
About this course
An IT capacity planning course that sizes one ticketing platform in a spreadsheet, for SREs, platform engineers and IT managers. You turn peak load and latency into requests in flight with Little's Law and check the worker slots, calculate the utilisation and response time of a single database writer with the M/M/1 model and see the hockey stick, compute p50, p95 and p99 latency with PERCENTILE.INC against an SLO, work out data transfer times in bits and bytes against a night window, forecast storage growth and the month to order capacity, size five services for the peak with 30% headroom and a spare and price them per 1,000 requests, combine serial and redundant tiers into one availability and its number of nines and find the throughput knee in a load test. The first lab is free.
Lessons
- Little's Law: Requests in Flight, Throughput & LatencyFree
- Utilisation & Queueing Delay: the M/M/1 Hockey Stick Pro
- Latency Percentiles: p50, p95, p99 & the Histogram Pro
- Bandwidth & Data Transfer Time: Bits, Bytes & Windows Pro
- Storage Growth Forecast: Compound Growth & Months Until Full Pro
- Servers for the Peak: Headroom, ROUNDUP & Cost per 1,000 Requests Pro
- Availability of Serial & Parallel Components Pro
- Load Test Results: Finding the Throughput Knee Pro