Vector Search easy
You are setting an SLO for a new ANN-backed recommendation search over 50 million vectors. A product manager asks for 100% recall@10 with single-digit millisecond latency. What is a realistic response?
- A Agree, since a well-tuned HNSW index always returns the exact top 10 at this scale and latency.
- B Agree, but only if you switch to IVF, which is exact while HNSW is approximate by design.
- C Target about 95-99% recall@10; reaching 100% means near-exhaustive search and higher latency.
- D Target about 50% recall@10, since ANN indexes cannot exceed that without exact flat search.
Answer: C. Target about 95-99% recall@10; reaching 100% means near-exhaustive search and higher latency.
ANN indexes trade recall for latency. The normal target is 95-99% recall@10, and pushing toward 100% forces the search to visit far more candidates. No ANN index, HNSW or IVF, guarantees exact results. If exactness is truly required, flat search is the answer, but at this scale it costs far more latency or hardware.
One system design question a day, with spaced repetition so the ones you miss come back.
Get the appRelated questions
- You are building an IVF index on 20 million vectors. A teammate asks why the build needs a 'training' step before any vectors can be added. What is the reason?
- An HNSW index serves a marketplace where about 30% of listings are updated or removed each week. Each update is applied as a delete plus an insert, and deletes mark tombstones. Over a month, recall@10 drifts from 97% to 89% and p99 latency rises at the same ef_search. What is happening, and what should you do?
- Users delete documents in your primary database, but the deleted documents keep appearing in vector search results for hours. The vector index is rebuilt nightly. What is the best fix?
- Your embedding model outputs unnormalized vectors whose magnitudes vary widely. A teammate wants to switch the index from cosine similarity to dot product because it is cheaper to compute. What is the consequence?
- A B2B SaaS app serves 500 customers, and every query searches only one customer's documents. One large customer runs heavy batch queries that slow down search for everyone. Which sharding approach fits best?
- Your HNSW index (M=16) reaches recall@10 of 91% at ef_search=64 with a p99 of 6 ms. The target is 97% recall within a 15 ms p99. A full rebuild takes 14 hours. What is the right tuning order?