Typorithm · easy

Scalability & Performance

scalability-basicssystem-designscalabilitylatencythroughput

Saved on this device

Full lesson · 8 learning steps Open any exercise for focused practice

Step 1 of 8 · intro card

Learn Scalability & Performance

Fundamentals

Latency is measured in percentiles (p50/p95/p99) while throughput is requests per second; availability is stated in nines. Vertical scaling means a bigger machine and horizontal scaling means more machines — and you always find the bottleneck before optimizing.

Latency and throughput are different things. Latency is how long one request takes, reported as percentiles like p50, p95, and p99 so the slow tail is visible; throughput is how many requests per second the system serves.

Availability is stated in nines — 99.9% (three nines) allows about 8.7 hours of downtime per year, 99.99% only about 52 minutes. More nines cost more redundancy.

Scaling has two axes: vertical means a bigger, more powerful machine, and horizontal means adding more machines behind a load balancer. Before you scale or optimize anything, find the bottleneck — the slowest stage — because that is what limits the whole system.

Latency = percentiles (p50/p95/p99); throughput = requests/sec.

Availability is measured in nines (99.9%, 99.99%).

Vertical = bigger machine; horizontal = more machines.

metrics:
  latency: p50-p95-p99
  throughput: requests-per-second
  availability: nines-99-99
  scaling: vertical-vs-horizontal
  bottleneck: find-and-remove

Read the implementation

Connect each part of the code to the concept before you type.

1. Latency percentiles

Latency is reported as percentiles — p50 is the median, p99 the slow tail — not as a single average.

latency: p50-p95-p99

2. Throughput

Throughput counts how many requests the system handles per second, independent of any one request's latency.

throughput: requests-per-second

3. Availability

Availability is quoted in nines — 99.99% means roughly 52 minutes of downtime a year.

availability: nines-99-99

4. Scaling axis

Vertical scaling buys a bigger machine; horizontal scaling adds more machines behind a load balancer.

scaling: vertical-vs-horizontal

5. Bottleneck

Always locate the bottleneck — the slowest stage — before optimizing, since that is what caps the whole system.

bottleneck: find-and-remove

A complete learning loop

Understand

Read the concept map and follow the code breakdown.

Reconstruct

Type the implementation, repair mistakes and work through every recall exercise.

Verify

Use the final check without hints; revisit weak concepts in short practice sessions.