Messaging Between Microservices

When creating microservices, don't forget about the design and movement of your data between the components!

Transcript

1. Messaging Between Microservices Lorna Mitchell, Developer Advocate, IBM
2. Microservices are components by a buzzier name @lornajane
3. Messaging is how data moves between the components @lornajane
4. Microservices or SOA @lornajane
5. Gathering Data @lornajane
6. A swarm of tiny apps talking to the same storage is not microservices; it is an HTTP-enabled monolith. @lornajane
7. Data Hygiene Access to each storage piece should be "owned" by one component • decoupling improves scalability • some platforms charge by execution time; don't wait for a network call • improves maintainability - and sanity @lornajane
8. Moving Data @lornajane
9. Start with Security We know how to secure HTTP connections, so do it! • Auth standards e.g. JWT • Security in transmission: use HTTPS @lornajane
10. One System, Many Pieces @lornajane
11. Consider the Medium Is this a web request, with headers? A message on a queue e.g. RabbitMQ or Kafka? @lornajane
12. Designing Data Formats • JSON or XML? • Binary formats? • Apache Thrift • Protocol Buffers • Avro • Data vs MetaData @lornajane
13. Ship the documentation first; it's the cheapest phase of the project to fix bugs in @lornajane
14. Docs-First API Design • Quick and cheap to produce and change • Everyone can participate and discuss • Details (including field names!) are sorted out early Documentation is the first deliverable, and it's a living artefact (keep it in your repo) @lornajane
15. Returning Data @lornajane
16. Components that work alone need excellent log arrangements. It is the only way they can communicate with you. @lornajane
17. Logging and Microservices • Standard, configurable logging setup • Use a trace_id to link requests between services • Aggregate logs to a central place, ensure search functionality • Collect metrics (invocations, execution time, error rates) • display metrics on a dashboard • have appropriate, configurable alerting @lornajane
18. Originator Feedback In a microservices application: • errors can occur at any time • the last microservice should send information to the originating application • remember that distributed systems get everything done but not necessarily in the order you expect @lornajane
19. Data In Microservices @lornajane
20. Data In Microservices • Where does data come from? Where does it enter the application? • How does data move safely around the application? • How is data conveyed back out of the application? @lornajane
21. Resources • Website: https://lornajane.net • Email me: lorna.mitchell@uk.ibm.com Further reading: • Designing Data-Intensive Applications By Martin Kleppmann • Production-Ready Microservices By Susan J. Fowler @lornajane