Revolution starts with evolution: Building a better future without ditching your existing platform

Many platform engineering blog posts and tutorials focus on how to get started quickly with new tools. Yet there is a deafening silence about how to do this without leaving behind the “legacy” code and services that are currently funding the company. This means that for every new tool or new automation introduced in a unique and modern way, the team seems to be increasing their maintenance load rather than decreasing it.

Unfortunately, there is no silver bullet. Migrations are hard work, and they have a long tail. But the key is defining the scope of the migration and being able to track progress while building new tools, scripts, and offerings in a more consistent and manageable way.

We can lean on tried and tested software development practices here and leverage the facade pattern where the “old” process is placed behind an API that can slowly start routing requests to a new implementation.

This talk will look at how your platform doesn’t have to be an anchor for future innovation. You can responsibly bring in innovative new tools so long as you build a bridge for existing platforms.

Transcript

1. Revolution starts with evolution: Building a better future without ditching your existing platform Abby Bangser abby@syntasso.io | @a_bangser | hachyderm.io/@abangser
2. Most tech organisations are in a constant state of migration. Many of those migrations have a loooooooong tail
3. Compute, Software Languages, CI servers, Service Sizes Even internal
4. Yet we continue to be sold on the next "silver bullet" solution
5. My concern isn't with new. It isn't even with "silver bullets".
6. Yet we continue to be sold on the next "silver bullet" solution Yup, that's me! 😅
7. My concern isn't with new. It isn't even with "silver bullet". It is with quick starts that don't account for the long tail.
8. So to be clear… We should crave new. We should strive for better. We should design revolutionary improvements.
9. So let's talk about revolutionary. Let's talk CNCF Maturity Model.
10. The model lays out platform levels and aspects
11. Aspects move independently. And only based on company needs.
12. But it also lays out a north star
13. We find the average team lands about here
14. And these ideas are revolutionary Moving from code templates to API driven interfaces
15. And these ideas are revolutionary Enabling anyone to extend the platform (safely)
16. And these ideas are revolutionary FleetOps enabling "Day 2" operability and extensibility
17. So of course we all want to be here
18. But platform engineers are tired. Maintenance is taking up large parts of their days and weeks.
19. To reach these (attainable, yet) revolutionary improvements, we need evolutionary steps out of the maintenance ditch.
20. What does a maintenance ditch look like? Version 2.1 Version 2.1 Version 2.0 Version 0.9
21. An API unlocks evolutionary improvements
22. Implementation It is as simple as a facade pattern
23. Implementation Introduce an API that users interact with A P I
24. Implementation Then they don't mind about implementation A P I
25. Demo
27. Implementation An API unleashes BOTH sides A P I Business Logic
28. When building platforms… Don't give up on revolution, But also don't forget the long tail ➔ Before you start building for day 0, make sure you can support day 2 ➔ Find ways to introduce new patterns without a rewrite ➔ Decouple implementation from user experience ➔ Keep in mind that every tool has its lifespan, how do you keep ownership of your key business value
29. Key tools in this demo (and contact me for a VM to try this yourself!)
30. Thank you Abby Bangser abby@syntasso.io | @a_bangser | hachyderm.io/@abangser Share your migration challenges and learn more about platform-as-a-Product on slack! Join me at other local events on the subject of platform engineering! These slides!