Microsoft

Which of these 2026 vector databases are credible at billion-vector scale? Select all that apply.

Define recall@10 for a vector index, and spell out exactly what 'ground truth' means in that definition.

At what scale does 'just put it in Postgres with pgvector' stop being the right answer?

Modern vector databases combine sparse (BM25) and dense vectors at the storage layer via two distinct patterns. Identify them.

Estimate the RAM cost of an HNSW index over 100M vectors at 1024 dimensions with M=32, including graph overhead.

Pre-filter versus post-filter strategies for metadata-constrained ANN search — what breaks at each extreme?

Aggressive metadata filtering can silently destroy recall on an HNSW index even when matches exist. Why?

What problem does DiskANN solve that HNSW does not, and when do you reach for it?

When does binary quantization beat Product Quantization as the production compression choice for a vector index?

Which deployment shape best meets a strict P99 TPOT SLA on bursty traffic?