← Lessons

quiz vs the machine

Gold1410

System Design

Data Replication Factor Tradeoffs

How many copies to keep balances durability, read throughput, cost, and write latency.

5 min read · core · beat Gold to climb

How Many Copies

The replication factor is how many copies of each piece of data the system keeps. The number is a balance, not a maximum, because every extra copy helps some properties and hurts others.

What More Copies Buy

  • Durability rises because more independent failures must coincide to lose data.
  • Read throughput rises because reads can spread across all copies.
  • Availability rises because a node loss still leaves live replicas.

What More Copies Cost

  • Storage cost scales linearly with the factor.
  • Write latency can rise if a write must reach a quorum of copies before acknowledging.
  • Consistency gets harder because more copies must be kept in agreement.

The Quorum Link

Replication pairs with quorum rules: with N copies, requiring W writes and R reads where W plus R exceeds N guarantees a read sees the latest write. Tuning N, W, and R lets a system slide between strong consistency and low latency.

Key idea

A higher replication factor buys durability, read throughput, and availability while costing storage and write latency, and quorum rules where W plus R exceeds N tie it to consistency.

Check yourself

Answer to earn rating on the learn ladder.

1. What does increasing the replication factor improve?

2. With N copies, what quorum condition guarantees a read sees the latest write?