System Design Interview Questions for Beginners

Deciding how the parts of a software system fit together, so that it meets its requirements for features, scale, reliability and cost.

Functional requirements say what a system does, such as sending a message. Non-functional requirements say how well it does it, such as latency, availability and scale.

Vertical scaling uses a bigger machine, and has a hard limit. Horizontal scaling adds machines, which needs stateless servers, but it scales further and gives redundancy.

It spreads requests across servers, removes unhealthy ones using health checks, and hides individual servers from clients.

Fast storage for a copy of data, which reduces latency and load on the database. Redis is a common example.

A network of servers around the world that keeps copies of files, so that users download from somewhere nearby.

When you have relationships, need joins and transactions, or have unpredictable queries. Choose NoSQL for a specific access pattern, such as simple key lookups at a huge scale.

Latency is how long one request takes. Throughput is how many requests are handled per second.

The share of time a system is working. 99.9% allows about 8.8 hours of downtime a year.

A part whose failure stops the whole system. Remove it with redundancy.

One that keeps no user data between requests, so that any server can handle any request. Sessions go in shared storage.

A structure, usually a B-tree, that speeds up lookups, at a small cost to writes and storage.

Keeping copies of data on several servers, for more read capacity and availability.

Splitting data across several databases, each holding a part, so that the system can handle more writes and more data.

A buffer between services. A producer adds work, and workers process it later, which makes responses faster and absorbs traffic spikes.

Authentication checks who you are, and authorisation checks what you are allowed to do.

A single entry point for clients, which handles routing, authentication and rate limits.

A monolith is one deployable application. Microservices are small services deployed separately, which helps large teams, but it costs more to run.

Restricting how many requests a client can make in a period, to stop abuse and protect services.

Clarify requirements, estimate scale, define the API and data model, then draw a simple design and improve it.