Skip to main content

Agnost Architecture Overview

Agnost is an open-source backend application development platform that operates on Kubernetes clusters, designed to be cloud provider agnostic, simplifying the switching process between cloud providers. The architecture of Agnost is composed of various technologies and components grouped into three logical clusters: Platform, Engine, and Cluster Resources.

High Level Architecture

Platform

The platform modules handle backend application design and development. This involves managing application data and enabling developers to create apps through Agnost Studio, the platform's front end. The key components are:

  • Platform-Core: The API server of the Agnost cluster. It handles user registration, organization, app and version creation, and manages data models, endpoints, cron jobs, message queues, and storage for each version.

  • Platform-Worker: Performs asynchronous tasks on behalf of the Platform-Core, such as sending emails.

  • Platform-Sync: A Socket.io real-time server used to send real-time messages about design and code changes of developed applications.

  • Platform-Studio: A React-based front-end application for managing settings, resources, and users of your Agnost cluster, and creating and deploying new applications.

Engine

The Engine modules manage application deployment and execution. Developed applications are deployed through the Agnost cluster Engine modules, which handle requests to an application endpoint. The main components are:

  • Engine-Worker: Handles the deployment of application designs to databases, API servers, and cron job schedulers. It also manages resources such as databases, caches, message brokers, and API servers in the cluster.

  • Engine-Monitor: Monitors and checks the health of all Agnost cluster resources, updating their status in the platform database, and sending real-time notifications in case of critical events.

  • Engine-Scheduler: Manages cron jobs defined in application versions, performing necessary updates and triggering cron job executions.

  • Engine-Realtime: A Socket.io server for apps that utilize real-time features of the platform.

  • API Servers: For each version of an app, there is an API server. These servers run the handler code for app endpoints, message queues, and cron jobs.

Cluster Resources

The Cluster Resources are the open-source technologies utilized by an Agnost cluster to develop, deploy, and run backend applications. These resources include:

  • Platform-Database: MongoDB is used to store all information about platform users and applications, serving as the single source of truth.

  • Platform-Cache: Redis is used to cache a subset of MongoDB data for faster access. It also caches user sessions, API keys, and uses its pub/sub mechanism to autoscale Socket.io servers.

  • Platform-Message Broker: RabbitMQ is used as a message broker to manage asynchronous task queues.

  • Cloud Storage: MinIO handles the document storage needs of the Agnost platform and the storage needs of the developed applications.

  • Knative: Allows the deployment of multiple versions of an application as Knative services, effectively utilizing cluster resources and automatically scaling workloads based on incoming request traffic.

These three layers - the Platform, Engine, and Cluster Resources - all work together to provide a robust, scalable, and user-friendly platform for backend application development. This architecture aims to simplify the backend development process, reduce infrastructure set-up time, and provide flexibility in terms of resource management and scalability. The beauty of Agnost lies in its integration of proven open-source technologies, providing core functionalities needed in almost every backend application out of the box.

For more information, suggestions, or comments about Agnost architecture, we encourage you to join our GitHub Repository discussions and become a part of the Agnost community.