Messing Up Your Microservices
- 1. Messing Up Your Microservices An Opinionated View The 100% Guaranteed Way Hamburg, 2022
- 2. 2 © Ginkgo Analytics Online Banking in 2019 • Business –> Buy now Pay Later • Employees –>1551 • Microservices –> 1500 Monzo Online Banking Messing Up Your Microservices
- 3. 3 © Ginkgo Analytics • Faster development cycles (easier deployment and debugging) • End to end ownership • Platform- and language agnostic services • Distributed architecture • Faster time to market • Better scalability • Standardized service contract • Loose coupling • Service abstraction • Service reusability • DRY principle (Don’t Repeat Yourself) • Service autonomy • Service statelessness • Service discoverability • Service composability • Greater agility • Isolated services have better fault tolerance • Easier to create a CI/CD pipeline for single- responsibility Messing Up Your Microservices Google Search Engine Promises
- 4. 4 © Ginkgo Analytics Let’s start Messing up Microservices Messing Up Your Microservices Architectural Tinkering Reinventing the Wheel Shiny Object Syndrome
- 5. 5 © Ginkgo Analytics Let’s adopt Shiny Object Syndrome Messing Up Your Microservices Psychological Culture Others •Job security and insurance policy •Justification of exceptional smart sounding title •Include the number of microservices rolled out in the team’s OKR •Micro-managers decide about the micro-services without architectural insights •Developing microservices without thinking about operational costs and ROI
- 6. 6 © Ginkgo Analytics Let’s Reinvent the Wheel – Why not? Messing Up Your Microservices Psychological Culture Others •Proud creator of service silos which might exist in the organization. •I hate microservices so I don’t care. •Everyone should make their own gateway service. •Want to use our service? Recall Separation of Concern. •Wrapper service around other service to make them secure and reliable •Let’s create our own configuration provider service
- 7. 7 © Ginkgo Analytics Let’s tinker Architectural Design • Build bottleneck services with synchronous APIs • Consider REST API as a microservice • Involve customer to assist in chaos testing because “Live Testing” is real thing. Design Thinking approach! • Too many services, let’s package them in a single place • Microservice across a horizontal layers e.g., data access or messaging. • Mocking Loose Coupling • We changed our service, so you better change as well otherwise we won’t talk • Abusing high functional cohesion • Bounded Contexts Mix-ups • Follow separation of concern, don‘t let anyone talk about your service, it is non of their concern • Adopt custom Logging mechanisms for micro ervices • ….. Messing Up Your Microservices
- 8. 8 © Ginkgo Analytics Design Bottleneck Service Messing Up Your Microservices
- 9. 9 © Ginkgo Analytics Design Bottleneck Service Messing Up Your Microservices
- 10. 10 © Ginkgo Analytics Design Bottleneck Service Messing Up Your Microservices
- 11. 11 © Ginkgo Analytics Mix Up Bounded Context Messing Up Your Microservices Microservices Monolith
- 12. 12 © Ginkgo Analytics Mix Up Bounded Context Messing Up Your Microservices Microservices Monolith
- 13. 13 © Ginkgo Analytics Mix Up Bounded Context Messing Up Your Microservices Service C Service B Service A Service F Service D Service H Service E Service I Service J Service G Microservices Monolith
- 14. 14 © Ginkgo Analytics Mix Up Bounded Context Messing Up Your Microservices Service C Service B Service A Service F Service D Service H Service E Service I Service J Service G Microservices Monolith Strangler Pattern? • Popular design pattern to incrementally transform your monolithic application into microservices. • It replacing a particular functionality with a new service. • Once the new functionality is ready, the old component is strangled. • The new service is put into use. • The old component is decommissioned altogether.
- 15. 15 © Ginkgo Analytics Mix Up Bounded Context Messing Up Your Microservices Service C Service B Service A Service F Service D Service H Service E Service I Service J Service G Microservices Monolith Strangler Pattern? • Popular design pattern to incrementally transform your monolithic application into microservices. • It replacing a particular functionality with a new service. • Once the new functionality is ready, the old component is strangled. • The new service is put into use. • The old component is decommissioned altogether.
- 16. 16 © Ginkgo Analytics Mix Up Bounded Context Messing Up Your Microservices Microservices Domain Driven Design –> Bounded Context • I need a mouse. Mouse in the context of computer hardware or Zoology • Will you take a pizza slice? Which and Why? • Similar to the decade long unsettled debate, which programming language is the best?
- 17. 17 © Ginkgo Analytics RESTful Service is a Microservice Messing Up Your Microservices REST Interface Database Cache Order Module Config Meta Fulfilment Module Shipping Payment
- 18. 18 © Ginkgo Analytics RESTful Service is a Microservice Messing Up Your Microservices REST Interface Database Cache Order Module Config Meta Fulfilment Module Shipping Payment
- 19. 19 © Ginkgo Analytics Messing Up Your Microservices Respect Tight Coupling – Cascading Failure Order Service Payment Service External PayPal Service Shipping Service Fulfilment Service Invoice Service Messaging Service
- 20. 20 © Ginkgo Analytics Messing Up Your Microservices Respect Tight Coupling – Cascading Failure Order Service Payment Service External PayPal Service Shipping Service Fulfilment Service Invoice Service Messaging Service
- 21. 21 © Ginkgo Analytics Messing Up Your Microservices Respect Tight Coupling – Cascading Failure Order Service Payment Service External PayPal Service Shipping Service Fulfilment Service Invoice Service Messaging Service
- 22. 22 © Ginkgo Analytics Messing Up Your Microservices Respect Tight Coupling – Cascading Failure Order Service Payment Service External PayPal Service Shipping Service Fulfilment Service Invoice Service Messaging Service
- 23. 23 © Ginkgo Analytics Messing Up Your Microservices Respect Tight Coupling – Cascading Failure Order Service Payment Service External PayPal Service Shipping Service Fulfilment Service Invoice Service Messaging Service
- 24. 24 © Ginkgo Analytics Summary • Microservices are not a silver bullet or Swiss Army Knife • Architectural decision must be made after thorough evaluation • Done wrongly, you will improve the throughput of 24/7 support team (daily fire fighting) • It may make your boss having some sleepless nights • Finally! Be creative in messing up Messing Up Your Microservices
- 25. HAMBURG - SHANGHAI – SINGAPUR - ZÜRICH Thanks! Do you have any Questions?