Hosting & Infrastructure

Overview

Hosting & Infrastructure covers the operational layer of Open edX — deployment tooling, database and cache configuration, Celery task queues, storage (S3), CDN, monitoring, and scaling. This is the domain of platform operators and DevOps engineers.

The Open edX community has standardized around Tutor (maintained by Edly/overhangio, not in the openedx org) for Docker-based deployment. Tutor replaced the earlier configuration repo (Ansible-based) as the primary deployment mechanism.

Current State (2026)

• Tutor: Docker Compose / Kubernetes-based deployment; the dominant deployment approach in the community; managed via tutor config and Tutor plugins
• Databases: MySQL for relational data; MongoDB for course content; Redis for Celery broker and caching; Elasticsearch for search
• Celery: Background task processing via Celery with Redis or RabbitMQ as broker; used for email, report generation, grade computation
• Storage: Static assets and user uploads typically stored in S3-compatible object storage
• CDN: Video and static assets served via CDN (CloudFront or equivalent)
• Monitoring: Prometheus + Grafana common; edx-django-utils provides DataDog/New Relic integration hooks

Architecture

• Service topology: LMS, Studio, course-discovery, credentials, forum, ecommerce (if used) as separate Django services; each with its own database
• Tutor orchestration: Tutor manages service configuration, networking, and deployment; plugins extend the base deployment
• Kubernetes: Tutor supports both tutor local (Docker Compose) and tutor k8s (Kubernetes) deployment modes
• Codejail: Isolated Python execution for graded problems; runs as a separate container/service with strict sandboxing

Relevant Repositories

Repository	Role in This Feature	Activity Level	Notes
openedx/openedx-platform	Django settings, wsgi, celery config	High	Main service
openedx/edx-django-utils	Monitoring, caching utilities	Medium	Operational tooling
openedx/codejail	Sandboxed Python execution	Low	Security isolation
openedx/codejail-service	Codejail as a microservice	Low	Containerized codejail
openedx/xqueue	External grader queue (legacy)	Low	For custom graders

Note: Tutor (overhangio/tutor) is not in the openedx org but is the de facto standard for deployment.

Recent Changes

• Tutor (maintained by Edly/overhangio) continues to be the community deployment standard
• configuration (Ansible-based) repo effectively deprecated

History

Origin

• Year introduced: 2012 (edX.org infrastructure was custom; community deployment was more ad-hoc)
• Initial implementation: Ansible-based configuration in the configuration repo; custom AWS infrastructure for edX.org
• Context: edX.org ran at massive scale; community deployments (xMOOC, national platforms) needed their own deployment tooling

Key Milestones

Year	Milestone	Teams / People Involved
2013	configuration repo (Ansible) open sourced	Unknown
~2019	Tutor project begins (by Régis Behmo / Edly)	Régis Behmo
~2021	Tutor becomes the recommended deployment approach	Unknown
~2022	configuration repo deprecated in favor of Tutor	Unknown

People Who Shaped This Area

• Engineering: Régis Behmo (Tutor creator, Edly); others unknown
• Product: Unknown — open question for interview
• Design: N/A (infrastructure area)

Open Questions

• [ ] What was the original edX.org infrastructure like (AWS services, AMI-based, etc.)?
• [ ] What drove the community to build Tutor rather than improving the Ansible configuration?
• [ ] How does the multi-service architecture affect operations complexity?
• [ ] What monitoring and alerting setups are used in production deployments?
• [ ] What are the most common infrastructure failures in Open edX deployments?