Evolution of edX LMS and Studio Architecture

The LMS and Studio architecture has evolved in many ways over the last few years and we are actively developing its future. This talk focuses on the past, present, and possible future of the edX-platform codebase.

Learn about how the edx-platform evolved as it quickly grew in team size, features, and usage since 2011. Be aware of the upcoming transformation of edX's technology-concerned monolith to a business-centric monolith. Additionally, learn how the edX architecture is scaling by investing in technology and design principles inspired by the Reactive Manifesto via asynchronous messaging/eventing, CQRS, and bounded contexts. These are illustrated through recent work on Block Transformers and the Grades architecture.

Transcript

1. May 26, 2017 Evolution of LMS and Studio Architecture Nimisha Asthagiri nimisha@edx.org github.com/nasthagiri
2. How have we evolved? Where are we now? What’s a desirable future? How are we getting there? What are some examples?
3. How have we evolved? Where are we now? What’s a desirable future? How are we getting there? What are some examples?
4. edx-pla(orm Commits
5. edx-pla(orm Contributors 0 10 20 30 40 50 60 70 edx-pla(orm contributors Trimester Total: 342
6. Team Growth 0 10 20 30 40 50 60 70 80 2013 2014 2015 2016 2017 edX Engineers year
7. File Growth 0 1000 2000 3000 4000 5000 6000 7000 Apr-13 Jun-13 Aug-13 Oct-13 Dec-13 Feb-14 Apr-14 Jun-14 Aug-14 Oct-14 Dec-14 Feb-15 Apr-15 Jun-15 Aug-15 Oct-15 Dec-15 Feb-16 Apr-16 Jun-16 Aug-16 Oct-16 Dec-16 Feb-17 Apr-17 Number of files .po files removed
14. Usage Growth 0 500 1000 1500 2000 2500 3000 3500 4000 4500 Number of enrollments in Thousands
15. Usage Growth 0 5 10 15 20 25 30 35 40 45 Compounded enrollments in Millions
16. Data Growth 0 500 1000 1500 2000 2500 3000 3500 4000 4/1/12 6/1/12 8/1/12 10/1/12 12/1/12 2/1/13 4/1/13 6/1/13 8/1/13 10/1/13 12/1/13 2/1/14 4/1/14 6/1/14 8/1/14 10/1/14 12/1/14 2/1/15 4/1/15 6/1/15 8/1/15 10/1/15 12/1/15 2/1/16 4/1/16 6/1/16 8/1/16 10/1/16 12/1/16 2/1/17 4/1/17 Number of courses enrolled courses total enrolled courses to date
17. How have we evolved? Where are we now? What’s a desirable future? How are we getting there? What are some examples?
25. How have we evolved? Where are we now? What’s a desirable future? How are we getting there? What are some examples?
26. Discoverability Extensibility Resilient ElasOc Responsive Modularity
27. ReacOve Manifesto ElasOc LMS: 13k rpm Studio: 60 rpm no bo;lenecks Responsive 99% < 2s consistent Resilient 99.9% upAme MTTF & MTTR isolaAon
28. How have we evolved? Where are we now? What’s a desirable future? How are we getting there? What are some examples?
29. ReacOve Manifesto ElasOc Responsive Resilient
30. ReacOve Manifesto ElasOc Responsive Resilient Message Driven event-driven asynchronous loosely coupled locaAon transparency
31. Asynchronous: OrchestraOon or Choreography?
32. Resources •  ReacOve Manifesto •  Self-contained Systems (SCS) •  Domain Driven Design, Eric Evans •  Building Microservices, Sam Newman •  Modularity, Parnas, 1972 •  Architecture Decision Records, Michael Nygard
33. Modularity Extensibility Main take aways •  Asynchronous non-blocking calls •  Separate reads from writes (CQRS) •  Bounded contexts •  Aligned best pracOces Responsive ElasOc Resilient Discoverability
35. Technical Concerns Steel Silver Wood PlasOc
36. Business Concerns Currency Hardware Kitchenware
37. Technical Concerns
38. Business Concerns
39. How have we evolved? Where are we now? What’s a desirable future? How are we getting there? What are some examples?
40. Recent examples Asynchronous Bounded Context CQRS Extensibility Grades ✔ ✔ ✔ Course sharing ✔ ✔ Course experiences ✔ ✔ Enterprise ✔ ✔ Course import/export ✔ Block transformers ✔ ✔ ✔ ✔ Course catalog ✔ ✔ Course overviews ✔ ✔ LTI consumer xBlock ✔ ✔
41. Block Transformers today’s max: 4,332 blocks in an MIT course
42. Block Transformers •  Collect phase for expensive operaOons –  Content tree traversal –  Denormalize –  xBlock instanOaOon via Modulestore –  Asynchronous, non-blocking •  Read phase for fast-access to block data –  User-specific and API-specific transformaOons –  Access read-opOmized data structure from AWS S3 –  Synchronous, blocking
43. Block Transformer ‘Collect’ Phase: Block Transformer ‘Transform’ (Read) Phase:
44. Example: Start Date Transformer ‘Collect’ Phase:
46. Best PracOces •  Join the discussions (OEPs, doc PRs, slack): –  Tests –  Code Structure –  Extensibility •  xBlocks + django •  Pluggable framework: Stevedore •  LTI + SSO •  Web hooks
47. Gracias Nimisha Asthagiri nimisha@edx.org github.com/nasthagiri
48. Extra Slides
49. Example: Content Tags Transformer ‘Collect’ Phase:
50. Data Scale 0 250 500 750 1000 1250 1500 1750 2000 2250 0-499 500 1000 1500 2000 2500 3000 3500 4000 number of courses number of course blocks max: 4332