Vector Search easy
You are building search over 200,000 internal documents with 768-dimension embeddings. Traffic is low, everything runs on one node, and the team wants exact results. Which index should you choose?
- A A flat brute-force index, since exact search over 200K vectors is fast enough on one node.
- B An HNSW index, because graph-based search is always the best choice for any vector workload.
- C An IVF index with product quantization to shrink the memory footprint as much as possible.
- D A sharded cluster partitioned by hash so that every query fans out across several machines.
Answer: A. A flat brute-force index, since exact search over 200K vectors is fast enough on one node.
Flat search is **exact**, needs no build or tuning, and stays fast up to about a million vectors on one node. HNSW adds memory overhead and build time, and it returns approximate results, so it buys nothing at this scale. HNSW becomes the right call at tens of millions of vectors or under strict latency at high QPS.
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?