Estimation
Back of Envelope.
The numbers every interviewer expects you to know — and a framework to use them under pressure.
Latency Numbers
Every latency number you need to know — L1 cache, RAM, SSD, HDD, network. Order of magnitude matters more than exact values.
L1/L2 CacheRAM vs SSDNetwork RTTJeff Dean's Numbers
Open topic →
Database Numbers
Read/write throughput, storage sizes, and query costs for SQL and NoSQL databases. What to expect at 1K, 10K, and 100K QPS.
Read ThroughputWrite ThroughputStorage SizesQuery Costs
Open topic →
Message Streams
Kafka throughput, partition sizing, consumer lag, and queue depth. Numbers for designing async pipelines that don't fall over.
Kafka ThroughputPartition SizingConsumer LagQueue Depth
Open topic →
Data Sizes
How big is a photo, a video, a user record, a log line? Wrong assumptions here break your storage estimates by 100×.
Media SizesRecord SizesStorage MathGrowth Rates
Open topic →
Bandwidth & Servers
Network bandwidth limits, server capacity, and how to translate QPS into CPU and memory requirements.
Bandwidth LimitsServer CapacityQPS to CPUMemory Math
Open topic →
Estimation Framework
A reusable step-by-step framework for any system design estimation. Structure your answer so the interviewer can follow along.
DAU to QPSStorage ProjectionServer CountInterview Structure
Open topic →