LEARN BY MAKING DECISIONS
From one request.
To a resilient system.
From a single request to a system ready to grow.
12 lessons / 12 experiments / Bilingual · Free · No sign-in
Trace the path
Step through how the system works
Change conditions
Tune parameters, inject failures, compare
Show understanding
Pass constraints and explain costs
Your learning path
201 minOne evolving service. A complete path through system design decisions. Basic HTTP and database knowledge helps. No distributed systems experience required.
Foundations
Trace a request and estimate exam-week peak traffic and concurrency.
Scaling
Calculate spare capacity after an API failure and how many database reads caching saves.
Data
Explain slow queries, stale reads, and hot spots using access paths, replica lag, and shard load.
Indexes
Run the same resource query as a full scan and an index lookup. Change table size and match ratio to compare the paths and work.
Replication
Update one resource from v1 to v2 and compare reads from the primary and replica. Adjust replication lag and wait time to see when stale reads end.
Sharding
Route the same batch of resource requests through one database or several shards in 3D. Adjust the shard count and hot request share to see which shard overloads.
Reliability
Estimate when a queue will drain, judge whether retries add pressure, and measure what users experience.
Queues
Move preview generation into background jobs and observe backlog growth, draining and duplicate deliveries.
Retries
Compare immediate retries, backoff with jitter and rate controls to keep partial failures from spreading.
Observability
Combine tail latency, errors, traffic and saturation to judge health using user-visible objectives.
Design Practice
Explain a choice during a network partition and propose a design that meets failure and budget constraints.
DESIGN IS A SERIES OF TRADEOFFS
Draw the system. Explain the decisions.
Finish with traffic, failure and budget constraints, then save your results and design decisions.
Explain requirements, estimates, tradeoffs, recovery and assumptions that need validation. Saved in this browser.