Progressive Delivery Talk at AWS re:Invent 2019

Our lightning talks at AWS this year were about how anyone can now adopt the Progressive Delivery practices of high-cadence Continuous Delivery leaders that have been in place for a decade or more at LinkedIn, Microsoft, Booking.com and others:
- Manage Gradual Releases by Decoupling Deploy From Release
- Monitor Gradual Releases to Limit The Blast Radius
- Experiment to Prove Features Have The Desired Impact

Split was founded by teammates that had used LinkedIn's XLNT/LIX platform (and others) and missed the power of such a platform once moving on to RelateIQ. After building an MVP at RelateIQ and once again seeing the power of such a platform, the founders set out to create a full-featured and robust commercial solution, extablishing Split in October of 2015.

Now the full power of a robust Feature Delivery Platform is available to anyone who subscribes to Split.

Full video of this talk (re-recorded in a quiet room away from the busy expo hall!), Q&A and resources for deeper learning are posted at https://www.split.io/blog/top-takeaways-aws-reinvent-2019/

Transcript

1. Trying to deliver features faster? Want to skip the hotfixes & rollbacks?
2. The future is already here — it's just not very evenly distributed. William Gibson
3. Progressive Delivery: - Control of Release - Observability of Impact At the user level
4. Split Feature Delivery Platform
5. Split Feature Delivery Platform
6. Manage is About Control of Exposure ...grant early access ...limit the blast radius ...define surface area for learning How do we decouple deploy from release? And decouple revert from rollback?
7. 7 Feature Flags for Gradual Rollout 0% 10% 20% 50% 100%
8. Split Feature Delivery Platform
9. Monitoring Quiz If you roll out a change to just 5% of your population and 20% (1 in 5) of those users get an error, what % of your total user population is getting that error?
10. 5% Exposure → ⅕ Get Errors = 1% Error Rate
11. Detect and Alert The Right Team
12. Split Feature Delivery Platform
13. 13 Feature Flags for Experimentation 50% 50%
14. Guardrail metrics
17. 3 Step Implementation: Create Split, Call SDK, Ingest Events Call Split SDK from your frontend, backend, or mobile apps. - In memory decisions - No private user data sent to Split 2 Ingest events via SDK, pre-built integrations, or open API 3 Create Split at app.split.io: - Split name - Targeting rules 1
18. Split Feature Delivery Platform