topic / performance-engineering

Performance Engineering

Measure the bottleneck, change one thing, and measure again.

Connect latency, throughput, concurrency, queues, profiling, tracing, CPU and memory behavior, query plans, tail latency, and load testing into a measurement process.

1 article · 1 video · 1 trace · 1 request flow · 1 interview · 5 total

Articles

performance engineeringSep 25, 2026

Measure the Work Before You Optimize It

Use workload shape, latency distributions, profiling, and controlled experiments to find the cause of slow software.

7 min readRead

Request Flows

request flow · 5 stagesA Performance Investigation Starts With a MeasurementFollow a controlled workload through baseline measurement, profiling, a targeted change, and a repeated comparison.

Videos

In production

performance engineeringSep 25, 2026

Diagnosing a Slow Service With Evidence

A planned investigation separates connection-pool wait, query time, application work, and response transfer before changing code.

Details

Traces

trace 01 · serverTrace Latency Waiting for a Database ConnectionFollow a request through a saturated connection pool and distinguish queue delay from database execution time.

Interviews

medium · t2Why Might a Service Be Slow When CPU Usage Is Low?A performance prompt about queues, connection pools, lock waits, network latency, tail behavior, and tracing the full request.

Explore adjacent

computer architecture computer systems databases backend engineering runtime execution latency throughput and queues profiling and benchmarking cache and query performance

New lessons by email

Get new explanations about performance engineering and related systems in your inbox.

One technical dispatch per week. No noise.