Functional CQRS Functional CQRS The double engine badass @Ramtop - - PowerPoint PPT Presentation

@Ramtop @Ramtop

Functional CQRS Functional CQRS

The double engine badass

@Ramtop @Ramtop

Have you ever

considered Event Sourcing architecture?

@Ramtop @Ramtop

CQRS: When and Why

Very Good for

✔ Complex mutable state ✔ Separation of logic ✔ Auditing required ✔ Orders, Trades, Editors

@Ramtop @Ramtop

CQRS: When and Why

Very Good for

✔ Complex mutable state ✔ Separation of logic ✔ Auditing required ✔ Orders, Trades, Editors

Not good for

✗ Read only aggregations (e.g. Reports) ✗ Stateless calculations (e.g. risk) ✗ Very High performance (e.g. exchange) ✗ Trivial CRUD (e.g. TODO lists)

@Ramtop @Ramtop

Implementation Gradient

Command Pattern Event Source CQRS Monolith CQRS Microservices

@Ramtop @Ramtop

Command pattern

✔ Allow replay to recreate State ✔ Separation of logic from infrastructure ✗ No automatic persistence ✗ Limited Scalability

@Ramtop @Ramtop

Event Source

✔ Allow replay ✔ Separation of logic ✔ Automatic persistence ✗ Limited Scalability

@Ramtop @Ramtop

Separation of Query Model

✔ Allow replay ✔ Separation of logic ✔ Automatic persistence ✔ High Scalability ✗ Complexity

@Ramtop @Ramtop

Split to Microservices

@Ramtop @Ramtop

Domain Driven Design

@Ramtop @Ramtop

Domain Driven Design

DDD is a creative collaboration between technical and domain experts. Domain language must be reflected in the code. Code must let the business intention show up

@Ramtop @Ramtop

Immut mutab abili lity Pur urity Hi Higher order fun unct ctions ns No exceptions Trans ansformat mations s whi which h prese serse se properties

Functional Programming

@Ramtop @Ramtop

DDD D + Even ent t Sourci Sourcing + Fun uncti tion

• nal

al St Style

There is no silver bullet but...

@Ramtop @Ramtop

@Ramtop @Ramtop

Protagonists

@Ramtop @Ramtop

What is a ADT?

Represent entities that can change state in an immutable world.

An algebraic data type is a kind of composite type, i.e. a type formed by combining other types. Wikipedia

@Ramtop @Ramtop

What is a Fold?

Specialized type of recursion

In functional programming, fold refers to a family of higher-

• rder functions that through use of a given combining operation,

recombine the results of recursively processing its constituent parts, building up a return value. Wikipedia

@Ramtop @Ramtop

Commands

A Command is a request for changing the internal state of the system A Command can “fail” if is not congruent with the current state of the System Each Command is executed in an atomic context

@Ramtop @Ramtop

Events

Events are the “atoms” of System state change Nothing can change without an event, every event can change only one transactional aggregate State + Event => State

@Ramtop @Ramtop

Values and Entities

They represent the state of the domain Values have no identity, Entities are distinct They are all immutables To keep state changes we have Algebraic Data Types

@Ramtop @Ramtop

Transactional Aggregates

An aggregate is an aggregation of Entities They are composed by Entities and Values It has all the information for a transaction unit Aggregates is what Events fold to

@Ramtop @Ramtop

Queries

A query ask for a snapshot of the state Queries typically need different data and denormalization from domain model. So we separate the models in CQRS. Queries are only eventually consistent with the domain

@Ramtop @Ramtop

Actors

More powerful and easy to use concurrency model than Threads/Locks They communicate asynchronously and potentially remotely Useful for Bounded Contexts and Services Remote Actors works like Microservices

@Ramtop @Ramtop

Event Store

Store tuples

(timestamp, type, uuid, event, version)

Query by index fast Dynamic query In memory implementation can be enough

@Ramtop @Ramtop

github.com/uberto/anticapizzeria

@Ramtop @Ramtop

Antica Pizzeria

• Real application is about finance products booking

but a pizzeria has a surprising similar domain

• We want to do implement the backend for a

ChatBot that will assist booking and enquires about orders.

@Ramtop @Ramtop

Events Tree

@Ramtop @Ramtop

Commands

@Ramtop @Ramtop

Events

@Ramtop @Ramtop

Queries

@Ramtop @Ramtop

Queries listen to Events

(t (to k keep ep u up wi with th the chan the changes ges)

@Ramtop @Ramtop

Let's fold Events

(t (to cr creat ate E Ent ntiti ties es)

@Ramtop @Ramtop

Entities as EventComposable

@Ramtop @Ramtop

Commands emit Events

@Ramtop @Ramtop

Logic is called by Commands

@Ramtop @Ramtop

Actors

(whe (where we e we can p can put al all thi this stu tuff)

@Ramtop @Ramtop

Application

@Ramtop @Ramtop

Tests can process C&Q

@Ramtop @Ramtop

Add a new State to represent Dispatch when a pizza left the shop. Keep track of delivery person.

@Ramtop @Ramtop

Live Code

@Ramtop @Ramtop

Will you consider Event Sourcing in the future?

@Ramtop @Ramtop

QA

Uberto Barbini @Ramtop github.com/uberto/anticapizzeria

@Ramtop @Ramtop