Master the ‘Three Highs’: Availability, Throughput, and Scalability in System Design

High Availability

High availability means achieving a high level of uptime, often expressed as "3 nines" (99.9%) or "4 nines" (99.99%), which allows only a few seconds of downtime per day. Redundancy is required, and common patterns include:

Hot‑Hot : Two instances receive the same input and both send output downstream; the downstream system must handle duplicate data.

Hot‑Warm : Only the Hot instance sends output; if it fails, the Warm instance takes over.

Single Leader : One leader instance receives data and replicates it to followers.

Leaderless : No leader; any write is replicated to other instances, and a quorum of reads/writes ensures data validity.

High Throughput

High throughput requires the service to process a large number of requests within a time window, measured by QPS (queries per second) or TPS (transactions per second). Techniques to achieve it include adding caches to avoid slow I/O, increasing thread counts for compute‑intensive tasks (while avoiding excessive threading), identifying bottlenecks, and using asynchronous processing to isolate slow components.

High Scalability

High scalability means the system can quickly expand to accommodate more capacity (horizontal scalability) or more functionality (vertical scalability). This is typically achieved by isolating responsibilities of each service, adopting micro‑services, and employing service discovery and load balancers to route requests to appropriate instances.

System Design Interviews

System design interviews are among the toughest technical interviews, requiring candidates to analyze vague problems, propose architectures that meet the three‑high goals, and communicate their ideas effectively. Success depends on understanding requirements, constraints, and bottlenecks, and applying the strategies described above.