Vector Search easy
Your product search team replaced exact search with an HNSW index. A QA engineer notices that the true closest item is occasionally missing from the top 10 results. What is the most likely explanation?
- A The HNSW graph is corrupted and must be rebuilt from scratch before results can be trusted.
- B The embedding model is non-deterministic, so the same query produces different vectors each call.
- C ANN indexes trade a small amount of recall for speed, so occasional misses are expected behavior.
- D The cosine metric is unstable for top-10 queries and should be replaced with exact L2 distance scoring.
Answer: C. ANN indexes trade a small amount of recall for speed, so occasional misses are expected behavior.
**Approximate** nearest neighbor search gives up a little recall (typically 95-99% recall@10) in exchange for much lower latency. You can raise recall by searching more of the graph (e.g., a larger ef_search), but latency goes up. Assuming corruption is tempting, but it is only worth investigating when recall falls far below the measured target or the misses follow a consistent pattern.
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?
- 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 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?