Dify
Dify is an open-source development platform designed for building, managing, and deploying applications powered by large language models (LLMs). Dify provides an intuitive interface that combines essential features such as AI workflow, Retrieval-Augmented Generation (RAG) pipelines, agent-based capabilities, model management, and observability tools, all of which help users transition quickly from prototype to production.
Key Features
AI Workflow Management: Streamlines the end-to-end development process for LLM apps.
RAG Pipelines and Agent Capabilities: Supports Retrieval-Augmented Generation and agent functionality to handle complex data queries and automate tasks.
Model Management: Enables efficient model management for various LLM deployments.
Observability: Provides tools to monitor and improve app performance in real time.
Installation and Deployment
You can deploy Dify using the Apolo, which facilitates Helm chart deployment and integrates with other applications running on Apolo, such as Postgres clusters, for seamless data handling.
The installation process automates:
Integration with existing PGVector app
Dify configs to persist datasets objects at Apolo Buckets
Web interface ingress configuration
Parameter Descriptions
api.replicas
Integer
Number of API service replicas (default: 1).
api.preset_name
String
Required. CPU/memory preset for the API service (e.g., cpu-small).
worker.replicas
Integer
Number of worker service replicas (default: 1).
worker.preset_name
String
Required. CPU/memory preset for worker service (e.g., cpu-small).
proxy.preset_name
String
Required. CPU/memory preset for the proxy service.
web.replicas
Integer
Number of web service replicas (default: 1).
web.preset_name
String
Required. CPU/memory preset for the web service.
redis.master.preset_name
String
Required. CPU/memory preset for Redis master instance.
externalPostgres.platformAppName
String
Required. PGVector app name to integrate with the Dify app.
externalPostgres.username
String
Optional. Username within the PGVector app (first available user used if not specified).
externalPostgres.dbName
String
Optional. Database to use within the PGVector app (first available DB used if not specified).
externalPgvector.platformAppName
String
Required. PGVector app name for direct vector data integration. Note: it could be the same externalPostgres app.
externalPgvector.username
String
Optional. Username for PGVector integration (first available user used if not specified).
externalPgvector.dbName
String
Optional. Database to use within PGVector for vector data storage.
This setup uses Helm chart for flexible and scalable deployment, enabling seamless integration with PostgreSQL and PGVector services. Adjust parameters based on your infrastructure requirements and previously installed Postgres app names.
After the installations, users could also integrate LLM inference and embeddings apps to this app. This part of documentation is under development
References
Last updated