Course contentsShow
← System Design
Lesson 564 of 1,91915. Consistency ModelsFree lesson

Common Quorum Configurations

Standard patterns like R=W=majority, R=1/W=N for read-heavy, and R=N/W=1 for write-heavy workloads.

Common Quorum Configurations

What you'll learn: How to configure R, W, and N values in quorum systems to optimize for different workload patterns.

The Core Idea

Once you understand the quorum condition (R + W > N), you need to choose actual numbers. Different configurations trade off between read performance, write performance, durability, and consistency guarantees. Three standard patterns emerge for common scenarios.

Standard Quorum Patterns

R=W=Majority (Balanced Workload)

Set R = W = ⌈(N+1)/2⌉ where N is typically 3 or 5.

Example: N=3, R=2, W=2

This balanced approach ensures strong consistency while distributing load. Both reads and writes hit a majority, so there's always overlap. Good for mixed workloads where neither operation dominates.

R=1, W=N (Read-Heavy Workload)

Only one replica needed for reads, but all replicas must acknowledge writes.

Example: N=3, R=1, W=3

Reads are lightning-fast since any replica can serve them. Writes are slow and less available (one failed node blocks writes). Perfect for configuration data, feature flags, or other read-dominated scenarios where writes are rare.

R=N, W=1 (Write-Heavy Workload)

Writes acknowledge from just one replica, but reads must contact all replicas.

Example: N=3, R=3, W=1

Writes complete quickly, but reads are slower and less available (one failed node blocks reads). Suitable for logging systems, metrics collection, or write-dominated workloads where you can tolerate slower reads.

The Trade-off Triangle

Think of it as adjusting sliders: improve read speed → sacrifice write speed (or durability). Improve write speed → sacrifice read speed (or consistency guarantees). The R + W > N constraint ensures you can't cheat consistency completely.

Key Takeaway: Choose R=W=majority for balanced loads, R=1/W=N for read-heavy systems prioritizing read speed, and R=N/W=1 for write-heavy systems prioritizing write throughput—each configuration trades off between read performance, write performance, and availability.