Roadmap › System Design — High-Level Design (HLD) from Basics to Case Studies › Module: Fundamentals
Lessons in this course (14)
2. Scalability, availability, latency and throughput
Basic★ Must knowLesson 2 of 14
| Term | Meaning |
|---|---|
| Scalability | Handle more load by adding resources |
| Vertical scaling (scale up) | Bigger machine — simple, limited, single point of failure |
| Horizontal scaling (scale out) | More machines — needs stateless services and data partitioning |
| Availability | % of time the system works; achieved by redundancy (no single point of failure) |
| Reliability | Works correctly (no data loss, correct results) |
| Latency | Time for one request (measure p50, p95, p99 — not the average) |
| Throughput | Requests handled per second |
| Durability | Data survives failures (replication, backups) |
Stateless services are the key to horizontal scaling: keep session/state in Redis, a database or the client token, so any instance can serve any request.
Availability in series and parallel: two components each 99.9 % in series → 99.8 %; two redundant copies of a 99 % component in parallel → 99.99 %.