A defining number
The read write ratio compares how many reads happen per write. Most consumer systems are heavily read dominated, often a hundred reads or more per write.
Why it steers design
- Read heavy systems benefit from caching and read replicas, which absorb reads cheaply.
- Write heavy systems need attention to write throughput, durable logs, and partitioning to spread writes.
A timeline that is read once per post but viewed thousands of times is a textbook read heavy case, so the design leans on caches and fan out.
Estimating it
- Count expected reads per user action and writes per action.
- Divide reads by writes for the ratio.
Misjudging this ratio leads to the wrong investment, such as building elaborate write scaling for a workload that is really starved for read capacity.
Key idea
The read write ratio tells you whether to optimize for cheap reads with caches and replicas or for high write throughput.