Services

A serverless service is a group of one or multiple functions and any resources they require. By grouping related functions together, it's easier to share code and and resources between those functions. Services are also designed to be completely independent, which helps teams develop more quickly, without waiting for others.

Each serverless service contains two configuration files which describe it:

serverless.yml
- Declares a serverless service
- Defines one or multiple functions in the service
- Defines events that trigger each function to execute (e.g., HTTP requests)
- Defines one set of resources (e.g., 1 AWS CloudFormation stack) required by the functions in this service
- Events listed in the events array may automatically create the resources required for the event upon deployment
- Config can be specified for one or more IaaS providers
- Re-usable and publicly shareable
- Contains no author-specific information
serverless.meta.yml
- Contains author-specific information (not intended for version control)
- Defines stages for this service
- Defines stage-specific variables, which allows adding dynamic values to serverless.yml, and helps keep out sensitive information
- The following variables are reserved: service, function, stage and region
- Specifies profiles or credentials to use per stage

Examples

AWS Example

Service Structure

users
  lib // contains logic 
  node_modules
  package.json
  serverless.yml
  serverless.meta.yml
  users.js // single handler file, requires lib

serverless.yml

service: users
description: A simple service for creating users

functions:
  create:
   extend: $${defaults}
   handler: users.create
   events:
     - http_endpoint_aws:
        path: users
        method: post

defaults:
  name_template: ${service}-${stage}-${function}
  include:
   - ../parent/data.json
  exclude:
   - .git
  aws_lambda_function:
    runtime: nodejs4.3
    timeout: 6
    memory_size: 1024

resources:
  - aws_name: ${service}-${stage}-resources
  - aws_description: Resources for the ${service} service in the ${project} project.
  - aws_dynamodb_table:
      name: ${service}-${stage}-users
      table_name: ${service}-${stage}-users
      provisioned_throughput: 1
      key_schema: id

plugins:
  - plugin
  - additional_plugin

serverless.meta.yml

 stages:
  dev:
   extend: $${defaults}
   creds:
    awsProfile: # stage specific profile
   vars:
    stageVariable: helloworld1
   regions:
    aws_useast1:
     creds:
      awsProfile: # optional, stage + region specific credentials
     vars:
      regionVariable: helloworld2

defaults:
  creds:
   awsProfile: default-profile
  vars:
   project: myApp

Deployment

These deployment steps always occur first:

The serverless.yml and serverless.meta.yml files are loaded into two objects in memory (e.g., service, meta)
If YAML is used, it's converted to JSON
References using Serverless variable syntax ${} or Serverless template syntax $${} are loaded
Loop through the resources property and collect resources for the targeted provider

Deployment On Amazon Web Services

If the targeted provider is AWS, and the serverless.yml contains AWS resources, these additional steps occur:

A default AWS CloudFormation template is loaded.
All of the resources in the resources property are added to that template.
The compute resources found in the functions are added to that template (e.g., AWS Lambda Functions).
Each event in the functions.yourFunction.events property is processed. If it requires resources, these are also added to the template.

Deployment On Microsoft Azure

Deployment On Google Cloud Platform

To be continued...

3.7 KiB Raw Blame History