Vector Search easy
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?
- A Shard by hash of document ID so every query is spread evenly across all nodes in the cluster.
- B Keep a single global index and add a tenant ID post-filter applied after each ANN search.
- C Shard by embedding dimension so that each node stores a different slice of every vector.
- D Shard by tenant so a heavy customer's load stays isolated and queries hit only one shard.
Answer: D. Shard by tenant so a heavy customer's load stays isolated and queries hit only one shard.
Sharding by **tenant** isolates noisy neighbors and avoids fanning every query out to every shard. The downside is uneven shard sizes when some tenants are much larger than others. Hash sharding spreads load evenly, but every query then touches all nodes, so the heavy tenant still slows everyone down. Hash sharding is the better choice for a single shared corpus with no tenant boundary.
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?
- 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?