Jon Rumsey

An online markdown blog and knowledge repository.

---

Project maintained by nojronatron Hosted on GitHub Pages — Theme by mattgraham

Serverless, Amplify and GraphQL

Resources

What is Serverless Architecture? By Hackernoon.com

Amazon AWS Amplify

Data Modeling at Amplify Docs

Serverless Architecture

A buzzword!

It is:

• An execution model: Cloud CSPs manage server allocation and provisioning.
• Stateless.
• Event-triggered.
• Priced based on execution count (rather than compute capacity, etc).
• Oriented to service client-side logic.
• A provider of Functions-as-a-Service.
• Back-end logic + Security + Database functionality. Client-side logic is... in the client.

Sample of CSPs and their Serverless offerings:

• IBM OpenWhisk
• AWS Lambda
• Azure Functions
• Google Cloud Functions
• Auto0 Webtask
• Oracle's Fn Project
• Kubeless
• Others

Down-sides:

• Neworking: Only accessed as private APIs.
• REQUIRED: An API Gateway (which the CSP probably provides for you, maybe at a cost).
• Larger services that are dependency heavy are not well suited to be ported to a serveless environment.
• Long-running processes (e.g. 300+ seconds) will get killed, so avoid implementing sevices that are longer-term and/or access external sevices in order to perform.
• Scaling might not have the configuration and capability you need (but maybe it will).

Functions As A Service (FaaS):

• Deploy individual Functions as components of larger business logic.
• Process individual requests within a timeout period.
• Invokation-based billint.
• Event-driven.
• "Instantaneously" scalable.
• Independent, Logical functions.
• Stateless.
• Ephemeral: Spin-up, do work, shut-down.
• Optional: Can be invoked manually.
• Managed entirely by CSP.

Serverless Solution Example:

• Web Server: Amazon S3 simple web server service.
• Lambda Function(s): Logging and data access (DB, JSON...) functions can be built as Functions.
• Security Token Service (STS): Temporary AWS credentials (API Key + Secret) for App users. Enables invoking Lambdas.
• User Authentication DB: AWS Cognito provides sign-up, sign-in to mobile and web apps. Allows OAuth or custom Ident Systems.
• Client App: UI, client-side rendering w/ JS and/or static HTML.
• Databse: AWS DynamoDB (NoSQL DB). Optional for serverless.

Common Serverless Frameworks and Architectures:

• Javascript, Python, Golang
• Apex: JS
• ClaudiaJS
• Sparta (Golang)
• Gordon (JS)
• Zappa (Python)
• Up (JS, Python, Golang, Crystal)

Links to Serverless Resource Materials

Serverless Examples

Anaibol Awesome-serverless

Build serverless contact form for static websites

Amazon AWS Amplify

Purpose-built tools and features.

FE Web and Mobile full-stack apps on AWS.

Configure web, mobile, backend for your app, visually, with content management.

Tools:

• Amplify Studio: Build full-stack App FE+BE visually.
• Amplify Libraries: Conenct App to existing services (Cognito, S3, etc)
• Amplify CLI: Manage Amplify from your local terminal.
• Amplify Hosting: Host website or content.

Use Cases:

• User Authentication: Plug-in your app to gain AuthN and user controls.
• Build Backend Data Model.
• Visually build App UI and Backend.
• Host progressive webapp or static website.

AWS Amplify Docs

Data Modeling With GraphQL

There are a bunch of code snippets throughout this article.

AWS Amplify automatically creates:

• Dynamo DB
• GraphQL types (@model directive in schema => @hasOne, @hasMany, @belongsTo, @manyToMany)

Setup Tables:

1. type Name @model { content: Type }
2. amplify push

Queries:

Queries look like a JSON-related plain text file with a heierarchy.

Example query:

query QueryAllThings {
  listThings() {
    things {
      items {
        id
        content
        createdAt
        updatedAt
      }
    }
  }
}
# Lists all 'things' (and their fields/properties?)

Primary Key:

Configure by using @model directive and id will be automatic and primary.

Mark a 2nd field with @primaryKey to specify it as the actual primary key.

Cannot change primary key without deleting/recreating DB table.

Sort Keys:

Use ID! without @primaryKey() directive.

Secondary Index:

Underlying datasource is DynamoDB, a KVP type NoSQL DB.

Model access patterns with secondary indices with @index directive.

Hash Key: Strict equality tester.

Sort Key: Comparisons including GT, LT, non-strict-equals, starts/ends with, and between operations.

sortKeyFields: ["fieldName"], is used within the @index directive.

Relationships Between Models:

Uni-directional 1:1 releationship between 2 models: @hasOne

One-direction 1:N relationship between 2 models: @hasMany

Has One or Has Many bi-directionsl relationship: @belongsTo

Join-Tables between 2 models for N:N relationships: @manyToMany

Note: Avoid circular relationships between uni-directional relationship tables, or uni- and many- relationship tables.

Bi-directional HasOne and HasMany relationships can be configured.

Default Field Values:

Use @default directive for scalar type fields e.g. int, string, etc.

An example from [Amplify Docs]:

type Todo @model {
  content: String @default(value: "My new Todo")
}

Rename and Disable:

Can be done to GraphQL queries, mutations, and supscriptions with name overrides, and nulling a value.

Other Comments:

• Custom queries can be created.
• Lambda functions can be used for queries.
• Edit timestamp fields 'updatedAt' and 'createdAt' to custom names like 'createdOn' and 'updatedOn'.

How It Works

See the bottom section of the site labeled How It Works for details on:

• Using the '@model' directive.
• TypeDefinition of '@model' directive.

Footer

Return to Root README