PRELIMINARY DRAFT

Definition

Goals:

1. Decouple from edx-platform

   • change and release forums independently of LMS, Studio etc

   • reduce complexity and size of edx-platform

   • move toward service-based ecosystem

2. Consolidate and standardize

   • merge entities and logic presently split across two separate code repos

   • use a single language

   • move away from special-case deployment (Heroku / add-ons)

3. Revisit data models and storage

   • data is split across three apps / database types

   • mongo is suboptimal fit for forums data model & access patterns; this causes a need for more application code/complexity to close the gaps

   • search index tightly coupled to mongo; search functions very limited

4. Improve product quality and extensibility

   • make it easier to understand and maintain the forums

   • make it easier to add key new features such as: moderation, ...

   • make it easier to operate and troubleshoot the forums (by eliminating opaque components and points of failure)

   • design for integration and interoperability: XBlock, LTI, alternate forums implementations (pluggability)

Constraints

1. Ensure continuity for past/present courses; do not interrupt service or data access for active courses in cutting over

   1. may have more flexibility with past / archived courses

   2. possibly start new courses on v2 while maintaining v1 for old?

2. Ensure continuity for research partners

   1. might be tricky, e.g. may be doing research over the span of a course, or comparing two courses assuming same version of forums, etc

   2. solve this going forward

3. Preserve baseline interoperability with alternate discussion systems (Piazza, SE, etc)

   1. maybe a non-issue - can alternate discusison providers be implemented with their own tabs?  more about how LMS links to other apps.

4. Be transparent to OSS community

5. Support / guide other platform deployments through a transition

Scope

User-Goal Matrix

Forums v2

Social Dialogue Software at MOOC Scale

Forums v2

Use cases (based on v1)

Integration Requirements

Platform (common)

• User model - identity, student profile, preferences

• Course-level constructs - course, course run, "commentable," cohort

• Course run lifecycle

• Course Enrollment events and course-level roles / access control

Studio

• automatically initialize forums for new course run

• create / modify / delete commentables

• inline thread preview

• export forum content in course run export

• import forum content with course run import

LMS

• discussion tab (desktop/browser app)

• inline thread view

• ajax endpoints (HTTP API)

Analytics

• emit student events

• provide bulk exports, partitioned by course or institution

LTI

TBD

XBlocks

TBD

System Requirements

Ops Requirements

UI and Accessibility Requirements

Capacity Requirements

Performance Requirements

Licensing Requirements

Relevant Business Targets (FY14)

• 180 Courses

• 2.3M Users

• 35 Universities + 5 B2B

Present (V1) Architecture

click for full-size view.

Build Versus Buy

Considering the nearly innumerable variety of third-party / off-the-shelf options for discussion forums products, why consider building?

Assuming that build was NOT an option, which of the existing products / competitors would be the best fit for us?  What is lacking from the best-fitting offering(s) that motivates us to choose our own implementation instead?

If we did choose to use a third party implementation, what impacts would that have on deployment, maintenance, and evolution of the rest of the platform?  On data integration?  Research capabilities?

(TBD: matrix in progress)