7. Performance

"Premature optimization is the root of all evil. But mature optimization is the root of all good."

What is Performance?

How efficiently a system uses resources to serve requests — measured in latency, throughput, and resource utilization.

Key Metrics

Latency

• p50 — Median (half of requests faster)
• p95 — 95th percentile (most users' experience)
• p99 — 99th percentile (tail latency, worst experience)
• p99.9 — Critical for high-traffic systems

Focus on percentiles, not averages. Averages hide the pain.

Throughput

• Requests per second (RPS)
• Transactions per second (TPS)
• Messages per second (MPS)

Resource Utilization

• CPU usage
• Memory usage
• Disk I/O
• Network bandwidth
• Connection pool usage

Key Concepts

Little's Law

L = λ × W

L = average number of items in system
λ = average arrival rate
W = average time in system

Amdahl's Law

Speedup = 1 / ((1 - P) + P/S)

P = proportion that can be parallelized
S = speedup of parallel portion

The serial portion limits total speedup.

Latency Numbers Every Programmer Should Know

L1 cache reference                  0.5 ns
L2 cache reference                    7 ns
Main memory reference               100 ns
SSD random read               150,000 ns (150 µs)
HDD random read            10,000,000 ns (10 ms)
Network round trip same DC    500,000 ns (500 µs)
Network round trip US→EU   150,000,000 ns (150 ms)

Topics

• ☐ Profiling techniques
• ☐ Database query optimization
• ☐ Caching strategies
• ☐ Connection pooling
• ☐ Async processing
• ☐ Compression
• ☐ CDN optimization
• ☐ Load testing
• ☐ Capacity planning
• ☐ Performance budgets

Optimization Process

1. Measure

• Profile before optimizing
• Identify bottlenecks with data
• Establish baseline metrics

2. Analyze

• Find the critical path
• Look for O(n²) operations
• Check for unnecessary work

3. Optimize

• Fix the biggest bottleneck first
• One change at a time
• Verify improvement with measurement

4. Validate

• Load test the change
• Check for regressions
• Monitor in production

Common Bottlenecks

Caching Strategy

Cache Hierarchy

Browser Cache → CDN → Application Cache → Database Cache → Database

Cache Patterns

• Cache-aside — App manages cache
• Read-through — Cache manages reads
• Write-through — Cache manages writes
• Write-behind — Async write to DB

Cache Invalidation

"There are only two hard things in CS: cache invalidation and naming things."

• TTL-based expiration
• Event-driven invalidation
• Version-based invalidation

Tools

Anti-Patterns

• Optimizing without measuring
• Caching without invalidation strategy
• N+1 queries
• Synchronous external calls
• Unbounded queries
• No connection pooling

Reading

• High Performance Browser Networking (Grigorik)
• Systems Performance (Gregg)
• Google SRE Book: Chapter 21 (Handling Overload)

Regulatory and control mappings

• ISO 27001 Annex A.8 Technological Controls A.8.6 Capacity management.
• ITIL 4 Practices Capacity and Performance Management.
• DORA ICT Risk Management Art 11 (performance under load + degradation).

Atoms

• USE Method for Resource Saturation

People-substrate cross-cluster

Sustainable human performance has the same shape as sustainable system performance: high throughput without burning out the substrate. CPTSD-fuelled performance is the human equivalent of running at 95% CPU -- looks great on the dashboard, breaks the moment load increases.

• Bridge essay: Interview Training as Applied Clinical Psychology -- Item 8 (over-indexing reads as virtue, indexes CPTSD)
• Psychology: Real emotional maturity -- the substrate that lets sustained high performance run without breaking the operator
• Psychology: Caretaker syndrome -- the anti-pattern: "high performance" via taking on everyone else's work; brittle, scales sub-linearly, ends in burnout
• Psychology: Hyper-independence -- the "if you want it done right" failure mode; locally fast, globally slow
• Competency: High Standards (IC) -- calibrated high standards = sustainable performance; over-indexed = perfectionism = brittle performance
• Anchor: Ownership psychology -- the calibration axis that distinguishes sustainable high performance from over-functioning that looks identical from outside