05Design patternsScaling reads
05 · Reusable pattern
Scaling reads
Layer replicas, caches, denormalized views, and CDNs while controlling staleness.Lesson spine
What you need to understand.
Read scaling is a ladder: optimize the query, add indexes, replicate, cache, precompute, and move reusable bytes outward only as pressure earns each step.
Optimize first
Measure the query plan, payload, serialization, and connection use before distributing the problem.
Read replicas
Send freshness-tolerant reads to followers and preserve read-your-writes with primary affinity or version waits.
Caching layers
Combine browser, CDN, service, and database caches only when each has a clear key and invalidation contract.
Denormalized read models
Precompute expensive joins or aggregates through versioned events when query latency matters more than immediate consistency.
Partitioned reads
Choose a shard key that keeps common queries local; scatter-gather raises tail latency and makes top-K merging expensive.
Protect hot data
Replicate hot immutable keys, coalesce fills, serve stale safely, and isolate pathological queries.
Before the boxes
Frame the decision.
What must work
Layer replicas, caches, denormalized views, and CDNs while controlling staleness.
What changes the design
Classify key lookup · range scan · aggregation · fan-out · payload · freshness
What owns the truth
Identify the component that commits authoritative state, then separate synchronous confirmation from derived work.
What stays simple
Do not add global coordination, multi-region writes, or a specialized store until a requirement earns the complexity.
Architecture map
Trace ownership, not just traffic.
Follow the decision from left to right. Every arrow should have a reason.
Commits canonical state
Propagates updates
Serves tolerant reads
Absorbs hot keys
Precomputes joins
Moves reusable bytes closer
Walk one representative request across every arrow. Say whether the handoff is synchronous or asynchronous, what identity makes a retry safe, and which step changes authoritative state.
- 01
Write source — Commits canonical state Define the output contract before moving to the next owner.
- 02
Change stream — Propagates updates Define the output contract before moving to the next owner.
- 03
Replica set — Serves tolerant reads Define the output contract before moving to the next owner.
- 04
Cache — Absorbs hot keys Define the output contract before moving to the next owner.
- 05
Read model — Precomputes joins Define the output contract before moving to the next owner.
- 06
Edge — Moves reusable bytes closer Confirm the result and emit the evidence needed to reconcile it.
Decision table
Make the trade-offs explicit.
| Decision | Defensible position | Cost to acknowledge |
|---|---|---|
| Primary mechanism | Every read optimization creates a freshness and invalidation contract. | The stronger guarantee usually adds coordination, latency, state, or operational work. |
| Sync vs. async | Keep only correctness-critical confirmation synchronous. Move derived views, notifications, analytics, and cleanup behind a durable boundary. | Async work needs idempotency, lag monitoring, replay, and a product definition for partial completion. |
| Simple vs. scaled | Begin with one logical owner and a clear API. Partition or replicate only the resource proven to be the first bottleneck. | Migration requires stable identities, versioned contracts, backfill, and a rollback path. |
Failure review
Design the recovery path.
Topic-specific risk
Replica lag and stale caches can violate read-your-writes expectations.
ResponsePersist enough identity and state to distinguish retry, resume, compensation, and operator repair.
Dependency timeout
A timeout is ambiguous: the remote side may have failed, succeeded, or still be running.
ResponseUse deadlines, bounded backoff with jitter, idempotency keys, and a status or reconciliation path.
Overload or skew
Average capacity can look healthy while a tenant, key, partition, region, or expensive request saturates one owner.
ResponseExpose queue depth and hot-key share, apply backpressure, isolate tenants, and degrade optional work before correctness.
Evidence + level bar
Prove the design can be operated.
Health of the promise
Measure user-visible latency or freshness, correctness drift, saturation, retry volume, and time to recover. Alert on the failed promise—not only CPU.
Complete and clear
Finish the happy path, identify the state owner, choose reasonable building blocks, and explain one scale mechanism.
Trade-offs and failure
Separate read and write paths, define consistency, explain partitioning, and make duplicate or partial failure safe.
Evolution and operations
Discuss multi-region boundaries, migration, tenant isolation, capacity, observability, and how the architecture changes over time.
Interview language
Open the deep dive with a claim.
“For Scaling reads, the decision I want to make explicit is this: Every read optimization creates a freshness and invalidation contract. I’ll trace the state-changing path first, show where the result becomes durable, then test the design against the highest-risk failure and our target scale.”
08 · Retrieval check
Can you defend it without the page?
- For Scaling reads, where is the correctness boundary and which failure would you test first?
- Which component owns committed truth, and what event or response proves the commit?
- Where is the first scaling or coordination bottleneck under the stated envelope?
- What happens after an ambiguous timeout or duplicate operation?
- Which complexity would you remove at one hundredth of the scale?