Prerequisites

You need the serverless framework installed globally
```
npm install -g serverless
```
Set your AWS credentials as environment variables using Isengard

Build

Run this before deploying or testing locally

npm install

Testing locally

sls invoke local --function api --path events/event.json

Deploy

Every time you call deploy, it changes an environment variable force it to redeploy each time. To disable comment out the following line in serverless.yml:

DEPLOYMENT_ID: Deployment-${sls:instanceId}

to comment use # at the beginning of the line

sls deploy

Switching between layer and esbuild

To switch between layer and esbuild

Comment out the layer in serverless.yml

#layers:
#  - !Sub 'arn:aws:lambda:${AWS::Region}:094274105915:layer:AWSLambdaPowertoolsTypeScript:21'

Comment out the exclude section in serverless.yml
```
#exclude:
 #- '@aws-lambda-powertools'
 #- 'aws-xray-sdk-core'
```
The code autodetects whether it is using a layer or not and sets a isLayer flag in the log, tracer annotation and metrics metadata.
Redeploy

What does the code do

The code imports the metrics, tracer and logger libraries and instantiates them. It creates an sts client and wraps it with the tracer. In the handler it calls getCallerIdentity to return some info to the caller and uses the metrics, tracer and logger libraries. Code

Benchmarking

If you don't want to manually deploy and look at the traces and logs, there are tools to help. The SDK blog post on reducing cold start times uses a modified version of this code for the benchmarking: https://github.com/lumigo-io/SAR-measure-cold-start It will change environment variables to force a cold start and invoke the function multiple times and report the results.

perpil / powertools-layer-repro