Edit on github

#AWS - Resources

If you are using AWS as a provider for your Service, all Resources are other AWS infrastructure resources which the AWS Lambda functions in your Service depend on, like AWS DynamoDB or AWS S3.

Using the Serverless Framework, you can define the infrastructure resources you need in serverless.yml, and easily deploy them.


Every serverless.yml using the aws provider is a single AWS CloudFormation stack. This is where your AWS Lambda functions and their event configurations are defined and it's how they are deployed. When you add resources those resources are added into your CloudFormation stack upon serverless deploy.

Define your AWS resources in a property titled resources. What goes in this property is raw CloudFormation template syntax, in YAML, like this:

# serverless.yml

service: usersCrud
provider: aws

resources:  // CloudFormation template syntax
      Type: AWS::DynamoDB::Table
        TableName: usersTable
          - AttributeName: email
            AttributeType: S
          - AttributeName: email
            KeyType: HASH
          ReadCapacityUnits: 1
          WriteCapacityUnits: 1

You can overwrite/attach any kind of resource to your CloudFormation stack. You can add Resources, Outputs or even overwrite the Description. You can also use Serverless Variables for sensitive data or reusable configuration in your resources templates. Please be cautious as overwriting existing parts of your CloudFormation stack might introduce unexpected behavior.

#AWS CloudFormation Resource Reference

To have consistent naming in the CloudFormation Templates that get deployed we use a standard pattern:

{Function Name}{Cloud Formation Resource Type}{Resource Name}{SequentialID or Random String}

  • Function Name - This is optional for Resources that should be recreated when the function name gets changed. Those resources are also called function bound
  • Cloud Formation Resource Type - E.g., S3Bucket
  • Resource Name - An identifier for the specific resource, e.g. for an S3 Bucket the configured bucket name.
  • SequentialID or Random String - For a few resources we need to add an optional sequential id or random string to identify them

All resource names that are deployed by Serverless have to follow this naming scheme. The only exception (for backwards compatibility reasons) is the S3 Bucket that is used to upload artifacts so they can be deployed to your function.

We're also using the term normalizedName or similar terms in this guide. This means dropping any characters that aren't allowed in resources names, e.g. special characters.

AWS Resource Name Template Example
S3::Bucket S3Bucket{normalizedBucketName} S3BucketMybucket
IAM::Role IamRoleLambdaExecution IamRoleLambdaExecution
Lambda::Function {normalizedFunctionName}LambdaFunction HelloLambdaFunction
Lambda::Version {normalizedFunctionName}LambdaVersion{sha256} HelloLambdaVersionr3pgoTvv1xT4E4NiCL6JG02fl6vIyi7OS1aW0FwAI
Logs::LogGroup {normalizedFunctionName}LogGroup HelloLogGroup
  • Schedule : {normalizedFunctionName}LambdaPermissionEventsRuleSchedule{index}
  • CloudWatch Event : {normalizedFunctionName}LambdaPermissionEventsRuleCloudWatchEvent{index}
  • IoT : {normalizedFunctionName}LambdaPermissionIotTopicRule{index}
  • S3 : {normalizedFunctionName}LambdaPermission{normalizedBucketName}S3
  • APIG : {normalizedFunctionName}LambdaPermissionApiGateway
  • SNS : {normalizedFunctionName}LambdaPermission{normalizedTopicName}SNS
  • Alexa Skill : {normalizedFunctionName}LambdaPermissionAlexaSkill
  • Schedule : HelloLambdaPermissionEventsRuleSchedule1
  • CloudWatch Event : HelloLambdaPermissionEventsRuleCloudWatchEvent1
  • IoT : HelloLambdaPermissionIotTopicRule1
  • S3 : HelloLambdaPermissionBucketS3
  • APIG : HelloLambdaPermissionApiGateway
  • SNS : HelloLambdaPermissionTopicSNS
  • Alexa Skill : HelloLambdaPermissionAlexaSkill
  • Schedule : {normalizedFuntionName}EventsRuleSchedule{SequentialID}
  • CloudWatch Event : {normalizedFuntionName}EventsRuleCloudWatchEvent{SequentialID}
  • Schedule : HelloEventsRuleSchedule1
  • CloudWatch Event : HelloEventsRuleCloudWatchEvent1
AWS::IoT::TopicRule {normalizedFuntionName}IotTopicRule{SequentialID} HelloIotTopicRule1
ApiGateway::RestApi ApiGatewayRestApi ApiGatewayRestApi
ApiGateway::Resource ApiGatewayResource{normalizedPath} ApiGatewayResourceUsers
ApiGateway::Method ApiGatewayMethod{normalizedPath}{normalizedMethod} ApiGatewayMethodUsersGet
ApiGateway::Authorizer {normalizedFunctionName}ApiGatewayAuthorizer HelloApiGatewayAuthorizer
ApiGateway::Deployment ApiGatewayDeployment{randomNumber} ApiGatewayDeployment12356789
ApiGateway::ApiKey ApiGatewayApiKey{SequentialID} ApiGatewayApiKey1
SNS::Topic SNSTopic{normalizedTopicName} SNSTopicSometopic
SNS::Subscription {normalizedFunctionName}SnsSubscription{normalizedTopicName} HelloSnsSubscriptionSomeTopic
  • DynamoDB: {normalizedFunctionName}EventSourceMappingDynamodb{tableName}
  • Kinesis: {normalizedFunctionName}EventSourceMappingKinesis{streamName}
  • DynamoDB: HelloLambdaEventSourceMappingDynamodbUsers
  • Kinesis: HelloLambdaEventSourceMappingKinesisMystream

#Override AWS CloudFormation Resource

You can override the specific CloudFormation resource to apply your own options. For example, if you want to set AWS::Logs::LogGroup retention time to 30 days, override it with above table's Name Template.

When you do overriding the basic resources, the important part is normalizedFunctionName. There are two rules.

  • Should start with uppercase character.
  • The - will be changed to Dash, _ will be changed to Underscore.

Here's an example:

    handler: handler.writePost
      - http:
          method: post
          path: ${self:service}/api/posts/new
          cors: true

        RetentionInDays: "30"