Vector Search easy
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?
- A IVF must fine-tune the embedding model on your data so the vectors become more accurate.
- B IVF must build a full HNSW graph first and uses training to pick the graph entry points.
- C IVF must learn coarse cluster centroids from a sample so each vector goes to a nearby list.
- D IVF must compute exact pairwise distances between all vectors to store them in a lookup table.
Answer: C. IVF must learn coarse cluster centroids from a sample so each vector goes to a nearby list.
IVF trains a **coarse quantizer** (centroids) on a representative sample. At query time it probes only a few lists, which trades some recall for speed and memory. Training never touches the embedding model, so option A is wrong. If the data distribution shifts significantly, the centroids go stale and the index should be retrained.
One system design question a day, with spaced repetition so the ones you miss come back.
Get the appRelated questions
- 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?
- 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?