Which approach is recommended to avoid index bloat with pgvector HNSW index on a table with frequent updates?
- A Rebuild index daily with REINDEX CONCURRENTLY
- B Batch updates and run VACUUM ANALYZE after (correct answer)
- C Use IVFFlat instead
- D Partition the table by date
Shortcut: Frequent updates can bloat HNSW indexes. Running VACUUM ANALYZE periodically removes dead tuples and updates planner statistics, maintaining query performance.