Skip to main content
Glama
docker-compose.licensed.yml4.2 kB
# Docker Compose for Licensed IRIS with ACORN=1 HNSW Optimization # FHIR AI Hackathon Kit - Enterprise Edition # Provides 10-50x faster vector search performance for multimodal medical data # # IRIS Enterprise Features: # - ACORN=1 HNSW optimization for vector search # - Enhanced ML capabilities # - Production-ready performance # # Adapted from rag-templates for FHIR-specific configuration services: iris-fhir-licensed: image: docker.iscinternal.com/intersystems/iris:2025.3.0EHAT.127.0-linux-arm64v8 container_name: iris-fhir-licensed ports: - "32782:1972" # IRIS SuperServer port (same as current iris-fhir) - "32783:52773" # IRIS Management Portal (same as current iris-fhir) environment: - IRISNAMESPACE=DEMO # FHIR data namespace - ISC_DEFAULT_PASSWORD=ISCDEMO # Match current FHIR setup volumes: - iris-fhir-licensed-data:/usr/irissys/mgr # Named volume for persistence - .:/home/irisowner/dev # Mount project directory - ./iris.key:/usr/irissys/mgr/iris.key # Enterprise license key stdin_open: true tty: true healthcheck: test: ["CMD", "/usr/irissys/bin/iris", "session", "iris", "-U%SYS", "##class(%SYSTEM.Process).CurrentDirectory()"] interval: 15s timeout: 10s retries: 5 start_period: 60s command: --check-caps false volumes: iris-fhir-licensed-data: {} # ============================================================================= # USAGE INSTRUCTIONS - FHIR AI Hackathon Kit (Licensed IRIS) # ============================================================================= # # 1. MIGRATION FROM COMMUNITY IRIS: # # a. Stop current iris-fhir container: # docker stop iris-fhir # # b. Export/backup current data (optional but recommended): # docker exec iris-fhir iris export /tmp/backup.gof DEMO # # c. Start licensed IRIS: # docker-compose -f docker-compose.licensed.yml up -d # # d. Wait for container to be healthy: # docker-compose -f docker-compose.licensed.yml ps # # e. Import data if backed up: # docker exec iris-fhir-licensed iris import /tmp/backup.gof DEMO # # 2. IRIS ACCESS: # - SuperServer: localhost:32782 (for Python apps) # - Management Portal: http://localhost:32783/csp/sys/UtilHome.csp # - Credentials: _SYSTEM / ISCDEMO # - Namespace: DEMO (FHIR data) # # 3. VERIFY ACORN=1 HNSW OPTIMIZATION: # - Access Management Portal: http://localhost:32783/csp/sys/UtilHome.csp # - Navigate to: System Administration > Configuration > Memory and Startup # - Check for ACORN=1 or iris_graph_core settings # - Run benchmark: python test_vector_search_performance.py # # 4. EXPECTED PERFORMANCE IMPROVEMENTS: # - Text vector search (3072-dim, 200K docs): <50ms (vs ~1-2s) # - Image vector search (1024-dim, 944 images): <10ms (vs ~100ms) # - Cross-modal search: <100ms (vs ~500ms) # - Multi-modal fusion: <200ms (vs ~1s) # - Throughput: 1000+ queries/sec (vs ~100 queries/sec) # # 5. DATA ALREADY IN IRIS (will be accessible after migration): # - VectorSearch.FHIRTextVectors: 199,969 radiology reports (OpenAI 3072-dim) # - VectorSearch.MIMICCXRImages: 944 chest X-rays (NV-CLIP 1024-dim) # - VectorSearch.FHIRResourceVectors: 51 clinical notes (384-dim) # - RAG.Entities: 171 medical entities # - RAG.EntityRelationships: 10 entity relationships # # 6. ROLLBACK TO COMMUNITY IRIS (if needed): # docker-compose -f docker-compose.licensed.yml down # docker start iris-fhir # # 7. MANAGING BOTH CONTAINERS: # - Licensed IRIS uses volume: iris-fhir-licensed-data # - Community IRIS uses volume: fhir-server_iris-fhir-data (or similar) # - Both can coexist but only one should run at a time (same ports) # # PORT MAPPING: # - Container port 1972 (SuperServer) → Host port 32782 # - Container port 52773 (Management Portal) → Host port 32783 # - Matches current iris-fhir setup for seamless migration # # ARCHITECTURE: # - Licensed IRIS 2025.3.0EHAT.127.0 (Enhanced Analytics Technology) # - ACORN=1 HNSW optimization for vector search # - Native VECTOR(DOUBLE, N) type support # - Production-ready for clinical decision support

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/isc-tdyar/medical-graphrag-assistant'

If you have feedback or need assistance with the MCP directory API, please join our Discord server