05Design patternsScaling writes
05 · Reusable pattern
Scaling writes
Partition ingestion, batch work, serialize conflicts, and absorb bursts without losing durability.Lesson spine
What you need to understand.
Write scaling moves from local efficiency to partitioning and buffering, then focuses on skew and coordination.
Efficient storage
Batch commits, use append-friendly logs or LSM trees, and keep indexes proportional to query value.
Horizontal partitioning
Hash for distribution, range for locality, or directory-based routing for control; plan resharding before the first hot shard.
Vertical partitioning
Separate immutable content, high-churn counters, and append-only events when their write shapes differ.
Queue and buffer
Absorb bursts behind a durable log, shape database load, and make queue age part of the product SLO.
Batch and aggregate
Combine increments or telemetry locally and hierarchically when the product accepts a freshness window.
Handle hot keys
Salt or split associative writes, assign a single writer, or isolate the celebrity key rather than sharding the entire store again.
Before the boxes
Frame the decision.
What must work
Partition ingestion, batch work, serialize conflicts, and absorb bursts without losing durability.
What changes the design
Peak mutation rate · bytes/write · key skew · contention · durability latency
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.
Creates mutations
Protects capacity
Chooses ownership
Orders accepted work
Applies durable state
Reclaims storage
Publishes query state
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
Producer — Creates mutations Define the output contract before moving to the next owner.
- 02
Admission — Protects capacity Define the output contract before moving to the next owner.
- 03
Partitioner — Chooses ownership Define the output contract before moving to the next owner.
- 04
Log — Orders accepted work Define the output contract before moving to the next owner.
- 05
Writer — Applies durable state Define the output contract before moving to the next owner.
- 06
Compactor — Reclaims storage Define the output contract before moving to the next owner.
- 07
Projection — Publishes query state 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 | Partitions add throughput until skew and cross-partition coordination dominate. | 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
A single hot key can serialize an otherwise healthy system.
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 writes, the decision I want to make explicit is this: Partitions add throughput until skew and cross-partition coordination dominate. 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 writes, 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?