What do I use if I need vector search plus keyword search in one system?

If you need both vector search and keyword search in one system, the common choice is a hybrid search engine.

Good options

• Elasticsearch / OpenSearch

  • Full-text keyword search
  • Vector similarity search
  • Hybrid ranking/filtering

• PostgreSQL with pgvector + full-text search

  • Good if you want to keep everything in one database
  • Works well for smaller to medium-scale systems

• Vespa

  • Strong hybrid search and ranking
  • More powerful, but more complex

• Milvus + a separate keyword engine

  • Usually not “one system” unless you combine components
  • Better for pure vector workloads

Best answer in practice

If you want one system for both, start with:

• OpenSearch or Elasticsearch for search-heavy applications
• Postgres + pgvector if you want simpler infrastructure and your scale is moderate

Rule of thumb

• Search app / enterprise docs / ecommerce → Elasticsearch/OpenSearch
• App with database-centered architecture → Postgres + pgvector
• Advanced ranking at scale → Vespa

If you want, I can also give you a decision table comparing these by scale, cost, and ease of use.