Deployment

To get the latest version of the Lambda function for execution, they should be deployed on the cloud. In this way, if our deployment package is heavier, then deployment will be slower. This will eventually degrade the performance of the function as it unpacks when it is invoked. The packing of our function also affects our execution. For example, if we are using Java, then packaging the function with all the dependencies grouped together as one function is slower compared to putting the dependencies within the lib folder. 

There are many tools that are available for deployment, but Serverless Framework is one of the main candidates for this job. Of course, there is lot that we still have to achieve, but this is still a far more mature solution than any other open source option available on the market. 

Also with respect to deployment , it should be orchestrated and should go through lots of approval and change request process with respect to UAT and prod environment. Deployment in production should be managed through service-now change request and CAB approvals.