05Design patternsSearch and full-text indexing
05 · Reusable pattern
Search and full-text indexing
Separate source-of-truth writes from asynchronous indexing, ranking, and retrieval.Lesson spine
What you need to understand.
Full-text search turns documents into analyzed tokens, distributed postings, and ranked candidates while filters narrow the work.
Inverted index
Store term-to-document postings with frequency and position so matching avoids full scans.
Analysis
Tokenization, normalization, language rules, synonyms, and stemming must be consistent at index and query time.
Distributed query
Route or scatter to shards, retrieve local top candidates, and merge scores at the coordinator.
Filters and facets
Execute selective exact filters early and use columnar doc values or dedicated structures for aggregations.
Ranking
Start with BM25, add field boosts and business signals carefully, and evaluate relevance with labeled queries and online outcomes.
Index synchronization
Use CDC or an outbox, carry source versions, and make update, delete, and replay monotonic.
Before the boxes
Frame the decision.
What must work
Separate source-of-truth writes from asynchronous indexing, ranking, and retrieval.
What changes the design
Corpus · terms/document · updates · fan-out · top-K · 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 records
Publishes versions
Analyzes fields
Maps terms
Builds plan
Collects candidates
Merges top results
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
Database — Commits records Define the output contract before moving to the next owner.
- 02
Outbox — Publishes versions Define the output contract before moving to the next owner.
- 03
Indexer — Analyzes fields Define the output contract before moving to the next owner.
- 04
Inverted index — Maps terms Define the output contract before moving to the next owner.
- 05
Query parser — Builds plan Define the output contract before moving to the next owner.
- 06
Shard fan-out — Collects candidates Define the output contract before moving to the next owner.
- 07
Ranker — Merges top results 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 | Document sharding simplifies updates; term sharding reduces some query work with harder balance. | 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
Out-of-order retries can resurrect deletes or replace newer documents.
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 Search and full-text indexing, the decision I want to make explicit is this: Document sharding simplifies updates; term sharding reduces some query work with harder balance. 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 Search and full-text indexing, 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?