need self hosted vector database on kubernetes

If you want a self-hosted vector database on Kubernetes, here are the most common solid options:

Popular choices

1. Milvus

   • Very popular for large-scale vector search
   • Good performance and features
   • More components, so deployment is a bit heavier

2. Weaviate

   • Easier to get started with
   • Supports vector search plus hybrid search
   • Good Kubernetes support

3. Qdrant

   • Lightweight and simple to operate
   • Strong for production use
   • Often easiest self-hosted option on K8s

4. pgvector on PostgreSQL

   • Best if you already use Postgres and your scale is moderate
   • Simpler stack, fewer moving parts
   • Not as specialized as Milvus/Qdrant/Weaviate for large-scale ANN search

---

Recommendation

• Small/medium production: Qdrant
• Hybrid search / easier developer experience: Weaviate
• Large scale / highest-end vector workloads: Milvus
• Already on Postgres: pgvector

---

Kubernetes deployment options

You can deploy these via:

• Helm charts (most common)
• Operators (best for production management if available)
• Raw manifests (least convenient)

---

Quick example: Qdrant on Kubernetes with Helm

helm repo add qdrant https://qdrant.github.io/qdrant-helm
helm repo update
helm install qdrant qdrant/qdrant \
  --namespace vector-db --create-namespace

---

Things to consider before choosing

• Vector count and expected growth
• Latency requirements
• Need for filtering / metadata
• Replication and persistence
• Operational complexity
• GPU requirement (usually not necessary for vector DB itself)

---

Persistence and scaling

Make sure your K8s setup includes:

• PersistentVolumes
• StorageClass with suitable performance
• Pod anti-affinity if running replicas
• Resource limits/requests
• Backups/snapshots

---

If you want, I can give you one of these:

1. a production-ready Helm install
2. a Kubernetes YAML example
3. a comparison table of Milvus vs Weaviate vs Qdrant
4. a recommended architecture for your workload