Telepresence Case Studies: From First Experience to Fast Feedback at Scale

Building and testing your microservice-based application becomes difficult when you can no longer run everything locally due to resource requirements. Moving to the cloud for testing is a no-brainer, but how do you synchronize your local changes against your remote Kubernetes environment? Following the usual container build-push-test cycle makes your inner development loop slower because you’d have to wait for minutes, sometimes hours, before seeing the impact of your code changes, and this dramatically reduces the number of iterations you can perform, the features you can ship to your end-users and negatively impacts the developer experience. Using three companies as a case study, this talk highlights how adopting the open source CNCF tool Telepresence improved their developer experience, accelerated their inner dev loop, and reduced staging environment compute costs. An explanation of what Telepresence is and a demo of how to implement it will also be covered in this talk.

Transcript

1. Telepresence Case Studies: From First Experience to Fast Feedback at Scale
2. Hello! My name is Edidiong Asikpo Senior Developer Advocate at Ambassador Labs. CNCF Ambassador. Enjoys writing, building, and sharing knowledge. Social media - @didicodes
3. Twitter: @didicodes Catch bugs before production Improve their developer experience Accelerate their inner dev loop
4. Twitter: @didicodes
5. Twitter: @didicodes
6. Twitter: @didicodes Things enables you to plan your day, manage your projects, and make real progress toward your goals.
7. Twitter: @didicodes • Use Kubernetes in AWS • Have multiple services • Use multiple managed databases Multiple microservices. Interdependently. Communicate. Process business requests.
8. Twitter: @didicodes Testing your local code changes against the other microservices and dependencies they rely on before deploying to prod is essential.
9. Twitter: @didicodes • Limited resources • Inconsistent data between production and non-production environments • Managing distinct configuration for separate environments This can be quite challenging!
10. Twitter: @didicodes The Problem? Inconsistency between their dev environment and production environment.
11. Twitter: @didicodes They wanted… A dev environment that allowed them to test their code changes against their external dependencies without the fear of things going wrong or not matching up after shipping to production.
12. Twitter: @didicodes Before Telepresence 😔 A development environment quite different from their production environment. With Telepresence 🎉 A development environment that resembles production very well. • “We can develop against the same database running in the exact same configuration.” • “When we develop something, we can be sure it’s 100% compatible with our production environment.” ▪ No surprising missing environment variable ▪ Incompatibility between production and mocked database ▪ Catch bugs before shipping to production
13. Twitter: @didicodes A collaborative conversation design platform where AI teams design, test, and ship chat and voice assistants faster.
14. Twitter: @didicodes Voiceflow was originally built as a monolithic application They eventually adopted the microservice architecture 20+ microservices, multiple databases, & queue engines
15. Twitter: @didicodes A tool that allows anyone to make voice apps without coding. • Applications became too large to run locally. • The development workflow was very slow. • Devs didn’t know what other devs were working on. • They couldn’t replicate the interdependency of cloud deployed apps locally, so their local testing wasn’t comprehensive enough. What challenges did Voiceflow face?
16. Twitter: @didicodes A tool that allows anyone to make voice apps without coding. • Applications became too large to run locally. • The development workflow was very slow. • Devs didn’t know what other devs were working on. • They couldn’t replicate the interdependency of cloud deployed apps locally, so their local testing wasn’t comprehensive enough. What challenges did Voiceflow face?
17. Twitter: @didicodes ~70 iterations of their code per day
18. Twitter: @didicodes Iterations per day drops from 70 to 40
19. Twitter: @didicodes The Problem? A slow inner development loop.
20. Twitter: @didicodes
21. Twitter: @didicodes “Other solutions introduce more complexity to their inner loop like adding file syncing with the cluster or checking changes in a Dockerfile, re-building it, and pushing it to the cluster.”
22. Twitter: @didicodes Before Telepresence 😔 With Telepresence 🎉 • The need to constantly build and deploy test images was removed • Increase in the speed & fidelity of their inner dev loop • They had to go through the build/push/test cycle before seeing the impact of their code changes • A slow inner dev loop
23. Twitter: @didicodes This is a large fintech organization that provides personal accounting services in the APAC region.
24. Twitter: @didicodes The Problem? Their Kubernetes development environment was negatively impacting their developer experience.
25. Twitter: @didicodes Cost Workstation Temperature Dev Loop Maintenance Cheap ✅ Fast 🚅 High ⛔ Hot 🔥 Local Development
26. Twitter: @didicodes Cost Workstation Temperature Dev Loop Maintenance Expensive 💰 Slow 🐢 Low ✅ Normal 🤩 Remote Development
27. Twitter: @didicodes Local Development Remote Development
28. Twitter: @didicodes Cost Workstation Temperature Dev Loop Maintenance Cheap ✅ Fast 🚅 Low ✅ Normal 🤩 Remocal Development
29. Twitter: @didicodes Without Telepresence 😔 A not so great developer experience With Telepresence 🎉 An improved developer experience • They bore the operational & resource burden of running all their microservices locally • Couldn’t utilize the benefits of both the “local” & “remote” K8s development methods • Code -> Build container -> Push to registry → Deploy & wait -> Test • Code -> Intercept ->Test • They got the best of both worlds (local and remote Kubernetes development) • The operational & resource burden of running all their microservices locally was removed
30. Twitter: @didicodes Catch bugs before production Improve their developer experience Accelerate their inner dev loop
31. Twitter: @didicodes What is Telepresence? • A tool that simplifies how teams develop and test on Kubernetes • Connects your local machine to a remote K8s cluster via a two-way proxying mechanism ▪ Access remote services as if they are running locally ▪ Reroute remote traffic to your local service • Seamlessly creates a remote-to-local “remocal” dev environment
32. Twitter: @didicodes Telepresence: How does it work?
33. Twitter: @didicodes Telepresence: Isolation through requests
34. Twitter: @didicodes Demo Video
35. Twitter: @didicodes Conclusion • Kubernetes & containerization has solved many challenges that business face today when it comes to flexibility, scaling & reliability of the release of new versions. • But it also posed additional challenges for developers by introducing operational complexity into their development workflow. • To solve this, we need to bridge the gap between remote K8 clusters and local development. • This is exactly what Telepresence does! It gives you the best of both worlds. • Try it out and/or contribute to this CNCF project.
36. Twitter: @didicodes Thank you! Questions? • GitHub - https://github.com/telepresenceio/telepresence • Product page - https://www.getambassador.io/products/telepresence • Telepresence for Docker - https://www.docker.com/products/telepresence-for-docker • Telepresence Pavillion - K30 • Ambassador Labs’ Booth - S7 • Slack channel - a8r.io/slack • Twitter - @ambassadorlabs