API First or Events First: Is it a Binary Choice?
1 like344 views
This document discusses the differences between APIs and events as architectural approaches and when each is best suited. It provides examples of when to use APIs versus events, including for business purposes like monetization, modernizing applications, and enabling new paradigms. The document also covers challenges of both approaches and shows how architectures can evolve from monolithic to microservices using a combination of APIs and events with loose coupling.
API First or Events First: Is it a Binary Choice?
- 1. Confidential │ ©2020 VMware, Inc. API First or Events First: Is it a Binary Choice? Madhav Sathe Cloud Applications Architect @madhav_sathe Last Updated: September 2020 Rohit Kelapure Solution Architect VMware @rkela APIs or Events ? What Came First ? https://miro.com/app/board/o9J_koawt4w=/
- 2. Confidential │ ©2020 VMware, Inc. 2 APIs and Events - The fundamental difference? Consumer API Do something for me APIs Events Producer Something happened Consumer I’m interested in it Pub/Sub Service Mesh and/or API Gateway
- 3. Confidential │ ©2020 VMware, Inc. 3 Your Digital Assets New Business Channels Internal Developers Partner Developers Customer Developers ERP, CRM API Gateway “By 2023, over 50% of B2B transactions will be performed through real-time APIs versus traditional approaches.” - Gartner, Use APIs to Modernize EDI for B2B Ecosystem Integration, 11 June 2019 ✓ Unlock new revenue streams ✓ Enable super infrastructure ✓ Adoption predicts financial performance Monetization through APIs Future growth engine Business Imperative - APIs ConsumptionExposure
- 4. Confidential │ ©2020 VMware, Inc. 4 ● Predictive analytics ● Fraud detection ● Real-time personalization ● IoT ● Easier to extend ● Autonomy to evolve independently ● Independence allows much better agility ● Serverless ● FaaS Business Imperative - Events Digital Experience → New Economic Opportunities Malleable Systems → Reduced Time to Market New Paradigms → Ride the Cloud Wave “By 2025, nearly 30 percent of data generated will be real-time” - IDC
- 5. Confidential │ ©2020 VMware, Inc. 5 Event Maturity Model Inspired By Martin Fowler
- 6. 6Confidential │ ©2020 VMware, Inc. Heuristics Choose Your Own Adventure
- 7. Confidential │ ©2020 VMware, Inc. 7
- 8. 8Confidential │ ©2020 VMware, Inc. “Successful software always gets changed” Fred Brooks
- 9. Confidential │ ©2020 VMware, Inc. 9 Monolith Evolution of Architecture Customer Module Order Module Customer Order
- 10. Confidential │ ©2020 VMware, Inc. 10 Microservices with tight coupling Evolution of Architecture Customer Service Order Service Customer Order A P I A P I
- 11. Confidential │ ©2020 VMware, Inc. 11 Microservices with loose but temporal coupling Evolution of Architecture Customer Service Order Service Customer A P I A P I Order
- 12. Confidential │ ©2020 VMware, Inc. 12 Decoupled microservices with thick events Evolution of Architecture Customer Service Order Service Customer A P I A P I Order Customer Events With State Order Events With State Order Query View Customer Query View
- 13. Confidential │ ©2020 VMware, Inc. 13 ● API’s externally observed behavior can not change unless it is backward compatible ● A good API practitioner will have sufficient tests to ensure backward compatibility ● Spring Cloud Contracts can be included in CI ● Use events with encoding standards that support schemas such as Avro ● Use schema registry to ensure compatibility ● Confluent’s schema registry maven plugin can be used in CI API & API Contracts Treat Event Schemas as API Contracts Managing Evolution Without Breaking Things Contracts & Compatibility
- 14. Confidential │ ©2020 VMware, Inc. 14 Design Considerations for Events Driven Event broker ● Delivery guarantees ● Message ordering Events ● Coupling due to event schema ● Message duplication State ● Sagas to handle eventual consistency ● Idempotency in operations Complete decoupling comes with a cost
- 15. Confidential │ ©2020 VMware, Inc. 15 API & Events Math API Events
- 16. Confidential │ ©2020 VMware, Inc. 16 Amundsen API Maturity Model API First Architecture Ease Of Consumption Product Cetricness
- 17. Confidential │ ©2020 VMware, Inc. 17 Patterns To Manage Uncertainty in Event Driven Architectures Reference Architectures
- 18. Confidential │ ©2020 VMware, Inc. 18 Docket Based Choreography Handling Uncertainty with APIs and Events @shaunanderson
- 19. Confidential │ ©2020 VMware, Inc. 19 Events Based Stateful Streaming Programming Models with Events @sina sujoodi
- 20. Confidential │ ©2020 VMware, Inc. 20 Exploit The Best Of Both Worlds Bringing It Together Start with APIs and API Gateways Immediate Impact Incremental Feature development Continue to decompose monoliths into a mesh of microservices Data Driven Strangler Model the System as Events Model Business Domain Completely Decoupled Microservices with Event Based Architecture and use APIs to drive value End Goal OPERATIONALIZE SEAMS EVENT DOMAIN MODEL
- 21. Confidential │ ©2020 VMware, Inc. Thank You
- 22. Confidential │ ©2020 VMware, Inc. 22 Tradeoffs and Functional choices Role of APIs and Events in the Design of A Complex System
- 23. Confidential │ ©2020 VMware, Inc. 23 ❖ Business ➢ Unlock new business channels by exposing data and services ➢ Montezing services ❖ Access ➢ Access across firewalls ➢ Endpoints in DMZ ➢ Abstract infrastructure and platform operations ➢ Platform & language neutral access to microservices ❖ Modernizing legacy ➢ Extending life of legacy with minimal disruption ➢ Anti-corruption layer ❖ Modern apps ➢ Backend for Frontend ➢ Temporal coupling is required ➢ Standards such as Service Mesh make it easy to implement zero-touch security When To Use APIs
- 24. Confidential │ ©2020 VMware, Inc. 24 ● Loosely Coupled Architecture ○ Temporal decoupling of microservices ○ Independent evolution of microservices ○ Extend architectures - Plug and play new modules ● Enables new Paradigms ○ Serverless microservices ● Data Modernization ○ Change data capture ○ Streaming & data transformation ○ Near real-time analytics ○ Modernizing ETLs ● Scaleable Business Processes ○ Implementing sagas ○ Notifications When To Use Events
- 25. Confidential │ ©2020 VMware, Inc. 25 ● Temporal coupling ● Needs extensive testing to ensure backward compatibility ● APIs Events Challenges with each approach
- 26. Confidential │ ©2020 VMware, Inc. 26 Decoupled microservices with thin events Evolution of Architecture Customer Service Order Service Customer A P I A P I Order Customer Events Order Events
- 27. Confidential │ ©2020 VMware, Inc. 27 API Maturity Model Comparison https://medium.com/good-api/api-maturity-fb25560151a3
- 28. Confidential │ ©2020 VMware, Inc. 28 Business Imperative - APIs & Events Path To App Modernization Reuse across multiple types of applications and integrations Enable Super Infrastructure - Forcing Function for Value Inclusive to humans & machines Adoption helps predict financial performance Make a system Malleable Easier to extend Greater autonomy to act independently Independence allows much better agility New paradigms - FaaS & Serverless API First Business Events First Business