In this chapter we’ll look at how our services will connect to each other while they are deployed across multiple environments.

Let’s quickly review the setup that we’ve created back in the Organizing services chapter.

  1. We have two repos — serverless-stack-demo-ext-resources and serverless-stack-demo-ext-api. One has our infrastructure specific resources, while the other has all our Lambda functions.
  2. The serverless-stack-demo-ext-resources repo is deployed a couple of long lived environments; like dev and prod.
  3. While, the serverless-stack-demo-ext-api will be deployed to a few ephemeral environments (like featureX that is connected to the dev environment), in addition to the long lived environments above.

But before we can deploy to an ephemeral environment like featureX, we need to figure out a way to let our services know which infrastructure environment they need to talk to.

We need our infrastructure resources and API services environments to be mapped according to this scheme.

API Resources
prod prod
dev dev
featureX dev
pr#12 dev
etc… dev

So we want all of the ephemeral stages in our API services to share the dev version of the resources.

Let’s look at how to do that.

In our serverless.common.yml (the part of the config that is shared across all our Serverless services), we are going to link to our resources app. You’ll recall that our resources are configured using CDK and deployed using SST.

First we start by defining the stage that our Serverless services are going to be deployed to.

  # Our stage is based on what is passed in when running serverless
  # commands. Or falls back to what we have set in the provider section.
  stage: ${opt:stage, self:provider.stage}

Next we create a simple mapping of the dev and prod stage names between our Serverless services and SST app.

  prod: prod
  dev: dev

This seems a bit redundant because the stage names we are using across the two repos are the same. But you might choose to call it dev in your Serverless services. And call it development in your SST app.

Next, we create the reference to our SST app.

sstApp: ${self:custom.sstAppMapping.${self:custom.stage},}-notes-ext-infra

Let’s look at this in detail.

  • First let’s understand the basic format, ${VARIABLE}-notes-ext-infra.

  • The notes-ext-infra is hardcoded to the name of our SST app. As listed in the sst.json in our resources repo.

  • The ${VARIABLE} format allows us to also specify a fallback. So in the case of ${VARIABLE_1, VARIABLE_2}, it’ll first try VARIABLE_1. If it doesn’t resolve then it’ll try VARIABLE_2.

  • So Serverless Framework will first try to resolve, self:custom.sstAppMapping.${self:custom.stage}. It’ll check if the stage we are currently deploying to (self:custom.stage) has a mapping set in self:custom.sstAppMapping. If it does, then it uses it. In other words, if we are currently deploying to dev or prod, then use the corresponding stage in our SST app.

  • If the stage we are currently deploying to does not have a corresponding stage in our SST app (not dev or prod), then we fallback to As in, we fallback to using the dev stage of our SST app.

This allows us to map our environments correctly across our Serverless Framework services and SST apps.

For reference, here’s what the top of our serverless.common.yml looks like:

  # Our stage is based on what is passed in when running serverless
  # commands. Or falls back to what we have set in the provider section.
  stage: ${opt:stage, self:provider.stage}
    prod: prod
    dev: dev
  sstApp: ${self:custom.sstAppMapping.${self:custom.stage},}-notes-ext-infra

Now we are going to use the resources based on the sstApp. Open up the serverless.yml file in the notes-api service.


custom: ${file(../../serverless.common.yml):custom}

    stage: ${self:custom.stage}
    tableName: !ImportValue ${self:custom.sstApp}-ExtTableName

The !ImportValue ${self:custom.sstApp}-ExtTableName line allows us to import the CloudFormation export from the appropriate stage of our SST app. In this case we are importing the name of the DynamoDB table that’s been created.

The provider: and environment: options allow us to add environment variables to our Lambda functions. Recall that we can access this via the process.env.tableName variable at runtime.

So in our list.js, we’ll read the tableName from the environment variable process.env.tableName.

const params = {
  TableName: process.env.tableName,
  KeyConditionExpression: "userId = :userId",
  ExpressionAttributeValues: {
    ":userId": event.requestContext.identity.cognitoIdentityId

The above setup ensures that even when we create numerous ephemeral environments for our API services, they’ll always connect back to the dev environment of our resources.

Next, let’s look at how to store secrets across our environments.