I'm unhappy with Milvus operational complexity

That’s a common pain point. Milvus can be powerful, but the operational overhead can feel heavy, especially once you’re managing dependencies, scaling, upgrades, and tuning.

If your main issue is operational complexity, the best alternatives depend on what you actually need:

If you want simpler ops with similar vector search

• Pinecone – fully managed, lowest operational burden
• Weaviate Cloud – managed option, easier than self-hosting
• Qdrant Cloud – simpler operational model, good search features

If you want to stay self-hosted but reduce complexity

• Qdrant – generally much simpler to run than Milvus
• pgvector on PostgreSQL – best if your vector workload is moderate and you want one system
• OpenSearch / Elasticsearch vector search – if you already operate those

If your use case is mostly application retrieval

• Postgres + pgvector is often the simplest reliable choice
• Works well when you don’t need massive scale or very specialized ANN tuning

A practical rule of thumb

• Need least ops effort: Pinecone
• Need self-hosted and simple: Qdrant
• Need “good enough” with existing DB stack: pgvector
• Need hybrid search / enterprise search: OpenSearch or Elasticsearch

If you want, I can help you choose a replacement based on:

1. data size
2. query latency target
3. budget
4. whether you need self-hosted or managed
5. whether you also need metadata filtering / hybrid search