TWJUG-Lite: Introduction to Event Sourcing

Essay of the talk: https://github.com/EddieChoCho/tech-talks-note/blob/master/2019/ChoChosBizarreAdventureOfEventSourcing.md

Youtube: https://youtu.be/7phuB-_ojOM

Transcript

1. ChoCho's Bizarre Adventure of Event-Sourcing A leisure study journey from 2017 to 2019 by Eddie Cho
2. Why I like to give this talk 1. Interesting 2. Issues that have to be overcome a. Fragmented study time b. The gap between concept and practices c. Found examples with unfamiliar languages/frameworks d. Stander implementation for Java + Spring + RDMBS Best Practices
3. Agenda 1. Core Concepts of Event Sourcing 2. Event Sourcing in Practice 3. Event Sourcing + Microservice
4. Core Concepts of Event Sourcing: Event Event: • An event is something that has happened in the past. • Conceptually, it should be Immutable. Event Sourcing is about: • Storing facts. • Reading information from the facts for different perceptions.
5. Core Concepts of Event Sourcing CRUD Table of Data Model (e.g. BankAccount) Id Owner Amount 9527 Arthur Morgan 500 W hy?
6. Core Concepts of Event Sourcing: Event Stream Reference: Event Sourcing in Practice - Benjamin Reitzammer & Johannes Seitz BankAccountCreated: {id: 9527, owner: Arthur Morgan} DepositPerformed: {accountId: 9527, amount: 1500} WithdrawalPerformed: {accountId: 9527, amount: 1000} DepositPerformed: {accountId: 9527, amount: 500}
7. Core Concepts of Event Sourcing: Event Store Type Content Aggregate Version BankAccountCreated {id: 9527, owner: Arthur Morgan} 9527 1 DepositPerformed {accountId: 9527, amount: 1500} 9527 2 WithdrawalPerformed {accountId: 9527, amount: 1000} 9527 3
8. Core Concepts of Event Sourcing: Projection Refernece: Event Sourcing in Practice - Benjamin Reitzammer & Johannes Seitz BankAccountCreated: {id: 9527, owner: Arthur Morgan} DepositPerformed: {accountId: 9527, amount: 1500} WithdrawalPerformed: {accountId: 9527, amount: 1000} BankAccount BankAccount {id: 9527, owner: Arthur Morgan} BankAccount {id: 9527, owner: Arthur Morgan amount: 1500} BankAccount {id: 9527, owner: Arthur Morgan amount: 500}
9. Core Concepts of Event Sourcing: Projection BankAccountCreated: {id: 9527, owner: Arthur Morgan} DepositPerformed: {accountId: 9527, amount: 1500} WithdrawalPerformed: {accountId: 9527, amount: 1000} BankAccountReport BankAccountReport {id: 9527} BankAccountReport {id: 9527, { totoalDepositAmount: 1500, totalWithdrwalAmount: 0, depositTimesPerMonth: 1 ….} BankAccountReport {id: 9527, { totoalDepositAmount: 1500, totalWithdrwalAmount: 1000, depositTimesPerMonth: 1 ….}
10. Core Concepts of Event Sourcing: Overview Refernece: Event Sourcing You are doing it wrong by David Schmitz Aggregate Events EventStream EventStore Projection Read Model
11. Core Concepts of Event Sourcing: Overview Issues could be solved by Event Sourcing 1. No audit log issue, Information would not be lost, Single source of truth 2. Structure changes more often than behavior => different read model 3. Allows you to time travel (Historic states) 4. Easier to anylize and fix issues. 5. Data analysis
12. Event Sourcing in Practice: Event Store Depends on your requirements, the event store could be... 1. eventstore.org (open source project) 2. NoSQL Database 3. Message Queue 4. RDBMS (?) 5. other options
13. Event Sourcing in Practice: Event Store RDBMS Event Store Design 1: OP lock with version Aggregate Table Event Table Reference https://cqrs.wordpress.com/documents/building-event-storage/ Column Name Column Type Id (PK) bigint Type varchar Version bigint CreateDate datetime Column Name Column Type Aggregste (FK) bigint Content blob(json) Version bigint CreateDate datetime
14. Event Sourcing in Practice: Event Store RDBMS Event Store Design 2: Event Table, ( id + parent should be unique) RDBMS Event Store Design 3: Use normal columns to replace context(blob) column Column Name Column Type Id (PK) bigint Content blob(json) Parent (FK) bigint CreateDate datetime
15. Event Sourcing in Practice: Snapshot Event 2 Event 1 ... Event 50 Event 51 Event 52 Snapshot 1 ~ 50
16. 1. Snapshots should not be places in line in the Event Log 2. Snapshots should have their own table 3. Memento pattern better insulates the domain over time as the structure of the domain objects change 4. Frequency of creating snapshots. <= Depends on your requirements Event Sourcing in Practice: Snapshot
17. Event Sourcing in Practice: Snapshot RDBMS Snapshot Table Design Reference https://cqrs.wordpress.com/documents/building-event-storage/ Column Name Column Type Aggregste (FK) bigint content blob Version bigint CreateDate datetime
18. Event Sourcing in Practice: CQRS CQS Command: void return type method, it is allowed to mutate state Query: non-void return type. it is not allowed to mutate state CQRS Command Component Query Component
19. Event Sourcing in Practice: Sample Project Event Sourcing + Spring Boot + MySQL https://github.com/EddieChoCho/event-sourcing-account-management Event Sourcing + Spring Boot + In Memory Event Store https://github.com/Pragmatists/eventsourcing-java-example/tree/projections
20. Event Sourcing + Microservice 1. Event-Driven Async Communication 2. CQRS 3. Choreography-based Saga Pattern Caution! These patterns are fit to be implemented with event sourcing. But they also could be implemented isolatedly.
21. Event Sourcing + Microservice: Async Communication Issue: communications with services through http request. 1. Tight coupling 2. Resilience Issue, Multiple points of failures. 3. Response time Order Service CoffeeBean Service CoffeeMaker Service Client
22. Event Sourcing + Microservice: Async Communication Solution: Event-Driven + Asynchronous Communication + Server Push Eventual Consistency Order Service CoffeeBean Service CoffeeMaker Service Message Queue Order Topic CoffeeBeen Topic CoffeeMaker Topic Client Command Query
23. Event Sourcing + Microservice: CQRS Reference: When Microservices Meet Event Sourcing Client Command Service Query Service Event Store Message queue Query DB (Read Models)
24. Event Sourcing + Microservice: CQRS Reference: Command Query Responsibility Segregation (CQRS) Order Service Query Service CoffeeBean Service CoffeeMaker Service Message Queue Order Topic CoffeeBeen Topic CoffeeMaker Topic
25. Event Sourcing + Microservice: Transaction Monolithic Service + RDBMS, ACID start transaction create order reserve credit commit
26. Event Sourcing + Microservice: Transaction Distribute transaction, Two-phase commit protocol 1. Synchronous communication 2. Resilience issue, Multiple points of failures. 3. Response time issue
27. Event Sourcing + Microservice: Transaction Choreography-based saga pattern: local transactions + publish messages (events) Order Service CoffeeBean Service CoffeeMaker Service Message Queue T1 T2 T3 T4
28. Event Sourcing + Microservice: Transaction Order Service CoffeeBean Service CoffeeMaker Service T1 C1 T2 Order Service CoffeeBean Service CoffeeMaker Service T1 C1 T2 T3 C2
29. Microservice + Event Sourcing: Sample Project Sebastian Daschner: Video Course — Event Sourcing, Distributed Systems & CQRS Event Sourcing + Microserives + Message Queue (Kafka) https://github.com/EddieChoCho/scalable-coffee-shop/tree/master/coffee-shop-blueprint
30. References Greg Young - CQRS and Event Sourcing The Many Meanings of Event-Driven Architecture • Martin Fowler Event Sourcing You are doing it wrong by David Schmitz Building an Event Storage - CQRS When Microservices Meet Event Sourcing Sebastian Daschner: Video Course — Event Sourcing, Distributed Systems & CQRS Event Sourcing in Practice - Benjamin Reitzammer & Johannes Seitz Microservice Patterns - Saga Pattern by Chris Richardson Greg Young — A Decade of DDD, CQRS, Event Sourcing
31. Further Readings Microsoft: CQRS Journey Patterns & Practices Symposium Online 2012 A Journey into CQRS CQRS Tutorial: http://www.cqrs.nu/tutorial/cs/01-design Leave feedbacks twjug slack: CCho twitter: @ChoCho008 Email: chochoforwork@gmail.com To Be Continued