Advanced GraphQL Architectures: Serverless Event Sourcing and CQRS @ ReactSummit

@slobodan_ AI version of Jorge Luis Borges (via ChatGPT) The
Labyrinth of Data About GraphQL

@slobodan_

@slobodan_ HTML HTML

@slobodan_ Mark Zuckerberg “Our biggest mistake was betting too much
on HTML5” 2012

@slobodan_ REST? FQL? REST? FQL?

@slobodan_ Lee Byron “We were frustrated with the di!erences between
the data we wanted to use in our apps and the server queries they required.” GraphQL: A data query language article, 2015

@slobodan_ Data app needs Data app needs

@slobodan_

@slobodan_ • Deﬁnes a data shape • Hierarchical • Strongly
typed • Protocol, not storage • Introspective • Version free

@slobodan_

@slobodan_ type Author { id: Int name: String posts: [Post]
} type Post { id: Int title: String text: String author: Author } Types

@slobodan_ type Query { getAuthor(id: Int): Author getPostsByTitle(titleContains: String): [Post]
} schema { query: Query } Schema

@slobodan_ getAuthor(_, args){ return sql.raw('SELECT * FROM authors WHERE id
= %s', args.id); } posts(author){ return request(`https://api.blog.io/by_author/${author.id}`); } Resolvers

@slobodan_ { getAuthor(id: 5){ name posts { title author {
# this will be the same as the name above name } } } } Queries

@slobodan_ • Parse • Validate • Run

@slobodan_ { "name": "Slobodan", "posts": [{ "title": "Hello World", "author":
{ "name": "Slobodan" } }, { "title": "GraphQL", "author": { "name": "Slobodan" } }] } Result

@slobodan_ GraphQL supports: • Queries • Mutations • Subscriptions

Advanced GraphQL Architectures Serverless Event Sourcing and CQRS

@slobodan_ AI version of Jorge Luis Borges (via ChatGPT) Computing,
without the machine About Serverless

@slobodan_ Ken Fromm “Why "e Future Of Software And Apps
Is Serverless” [link] 2012

@slobodan_ AWS announced AWS Lambda in 2014, and API Gateway
in 2015

@slobodan_ Gojko Adzic “It is serverless the same way WiFi
is wireless” Serverless architectures: game-changer or a recycled fad?, 2016

@slobodan_ What's serverless good for? • Event-driven apps • Teams
that are moving fast • Spiky workloads • Most of the apps

@slobodan_ What's serverless not good for? • Amazon Prime Video
• Basecamp • Things like long-running processes, ultra-low latency, or speciﬁc hardware or data geo-restrictions

@slobodan_ Serverless was (and still is) also great for prototypes

@slobodan_ A prototype

@slobodan_ Slobodan Stojanović CTO and co-founder of Vacation Tracker co-author
of Serverless Apps with Node.js book AWS Serverless Hero JS Belgrade meetup organizer

Our prototype became a real app

Our prototype problems became problems of a real app

Luckily, an old solution is waiting for (almost) all new
problems

@slobodan_ AI version of Jorge Luis Borges (via ChatGPT) Narrating
data's timeless journey About Event Sourcing

Remember Redux?

Deposit: $100000 Balance: $100000

Deposit: $100000 Pay: $2500 Balance: $97500

Deposit: $100000 Pay: $2500 Withdrawn: $5000 Balance: $92500

On a high level Redux has the same idea as
Event Sourcing

@slobodan_ Martin Fowler “Event Sourcing ensures that all changes to
application state are stored as a sequence of events.” Event Sourcing, 2005

@slobodan_ Martin Fowler “Not just can we query these events,
we can also use the event log to reconstruct past states” Event Sourcing, 2005

@slobodan_ AI version of Jorge Luis Borges (via ChatGPT) Two
sides, one truth About CQRS

@slobodan_ Bertrand Meyer “command query separation” Object-Oriented Software Construction book,
1988

@slobodan_ Martin Fowler CommandQuerySeparation, 2005

@slobodan_ Command and Query Responsibility Segregation (CQRS) Greg Young 2010

@slobodan_ Greg Young “CQRS uses the same deﬁnition of CQS
that Meyer used and maintains the viewpoint that they should be pure.” CQRS Documents, 2010

@slobodan_ Greg Young “"e fundamental di!erence is that in CQRS
objects are split into two objects, one containing the Commands one containing the Queries.” CQRS Documents, 2010

@slobodan_ How CQRS works? Vaughn Vernon Really Simple CQRS, 2018

Eventual Consistency

CQRS is an architecture pattern

You can, and most likely should, use CQRS in a
speciﬁc portion of your application.

@slobodan_ Martin Fowler CommandQuerySeparation, 2005

@slobodan_ CQRS GraphQL

@slobodan_ CQRS Query Command GraphQL

@slobodan_ CQRS Query Command GraphQL Query Subscription Mutation

@slobodan_ AI version of Jorge Luis Borges (via ChatGPT) Inﬁnite
queries, endless possibilities About Serverless CQRS with GraphQL

@slobodan_ Commands

@slobodan_ Queries

@slobodan_ Bonus Points: Subscriptions

Let's try to build this using Amazon Web Services (AWS)

And let's try to make it serverless

Step 1: Managed GraphQL AWS AppSync

@slobodan_ AppSync • Managed and scalable GraphQL • Out-of-the-box subscriptions
• Integrates with many services • VTL and JavaScript resolvers are built-in

Step 2: Events and projections storage Amazon DynamoDB

@slobodan_ DynamoDB • Managed and scalable NoSQL DB • Low
latency • Pay-per-use pricing • Can handle millions of requests per second • Can stream the data

Step 3: Create projections AWS Lambda

@slobodan_ Lambda • Managed and scalable functions • Cheap •
Pay-per-use pricing (1 ms) • Many natively supported languages + containers

@slobodan_ Commands & Subscriptions

@slobodan_ Queries

Step 4: Event Bus AWS EventBridge

@slobodan_ EventBridge • Managed and scalable event bus • Custom
events • Many integrations • Event storage and replay • Event schema & schema discovery

@slobodan_ Queries

Step 5: A Queue Amazon SQS

@slobodan_ SQS • Fully managed • Simple and highly scalable
• Proper queue with FIFO mode, retries, etc.

Let's see this in practice

1. The client sends an API POST request or GraphQL
mutation 2. The event is stored in the Events table (append-only, no edits) 3. The DynamoDB streams the event to the Lambda function that sends it to the EventBridge event bus 4. EventBridge triggers the specific business logic Lambda function 5. The business logic Lambda stores the "cached" data to one of the read-only DynamoDB tables 6. And then triggers the mutation that sends a "fake" mutation 7. A "fake" mutation triggers the GraphQl subscription to notify the clients

1. The client sends an API POST request or GraphQL
mutation 2. The event is stored in the Events table (append-only, no edits) 3. The DynamoDB streams the event to the Lambda function that sends it to the EventBridge event bus 4. EventBridge triggers the specific business logic Lambda function 5. The business logic Lambda stores the "cached" data to one of the read-only DynamoDB tables 6. And then triggers the mutation that sends a "fake" mutation 7. A "fake" mutation triggers the GraphQl subscription to notify the clients 8. App uses an event bus to "route" the response to the user's platform

Client then query the read-only tables directly using GraphQl

Client then query the read-only tables directly using GraphQl Or
using the RESTful API in case of bots

@slobodan_ Benefits •Fully Managed GraphQL •Easy to understand what happened,
when and why •Less code because of the resolvers •Better control •Highly scalable, but still cheap

@slobodan_ Downsides •Many new AWS services to learn •Harder to
trace the events •Increased complexity of the app •Hard to plan and update DynamoDB single table design

@slobodan_ Modeling data

Event storage: DynamoDB single table design

DynamoDB single table design Use ChatGPT as an assistant! See
Day 17 on knowlo.co/blog/ https://knowlo.co/blog/day-17-building-an-mvp-modeling-the-database/

Event schema: TypeScript

@slobodan_ Challenges & Solutions •Development & testing •Monitoring & tracability
•Security •Onboarding new team members •Evolving the data model and backwards compatibility Each of these bullet points can be a separate talk

@slobodan_ Environments & Cost

@slobodan_ Our environments 12 similar environments total cost: $1250/month

@slobodan_ Instead of a summary

@slobodan_ Stephen Wolfram “"e thing that has been sort of
a story of my life is realizing that simple rules can do much more complicated things than you imagine.” on Lex Fridman podcast

@slobodan_ Stephen Wolfram “"at something that starts simple and starts
simple to describe, can grow a thing that is vastly more complicated than you can imagine.” on Lex Fridman podcast

@slobodan_ Instead of a summary #2

@slobodan_ Joe Armstrong “Make it work, then make it beautiful,
then if you really, really have to, make it fast.” creator of Erlang programming language

@slobodan_ Joe Armstrong “90% of the time, if you make
it beautiful, it will already be fast. So really, just make it beautiful!” creator of Erlang programming language

@slobodan_ #CofounderGPT & https://knowlo.co/blog

Thank you! twitter: @slobodan_

Advanced GraphQL Architectures: Serverless Even...

Advanced GraphQL Architectures: Serverless Event Sourcing and CQRS @ ReactSummit

More Decks by Slobodan Stojanović

Other Decks in Programming

Featured

Transcript