Qdrant

Qdrant provides vector similarity search for the AI service, storing schema embeddings, document embeddings, and semantic search indexes.

Service Connection

# From ai-service values
config:
  qdrant:
    host: "qdrant.matih-data-plane.svc.cluster.local"
    port: 6333
    collectionName: "schema_embeddings"
    vectorSize: 1536

Collections

Collection	Vector Size	Purpose
schema_embeddings	1536	Database schema vectors for NLP-to-SQL
document_embeddings	1536	Document context for RAG
query_patterns	768	Historical query pattern matching

Deployment

Qdrant runs as a StatefulSet with SSD storage for fast vector lookups:

resources:
  requests:
    cpu: "500m"
    memory: "2Gi"
  limits:
    cpu: "2"
    memory: "8Gi"
 
persistence:
  enabled: true
  storageClass: ssd
  size: 50Gi

Protocols

Protocol	Port	Purpose
HTTP/REST	6333	Vector CRUD and search
gRPC	6334	High-performance operations

Dgraph Neo4j