Friday, June 16, 2017

AWS Lambda with Serverless framework

Google CloudFunction, IBM OpenWhisk, Azure Functions and AWS Lambda allow building applications in a Serverless fashion. Serverless doesn't mean that there is no server, but the Cloud vendor takes care of the provisioning the servers, scaling them up and down as required. In the Serverless model all we have to do is author a function with the event that triggers and the function and the resources the function uses.

Since, no servers are being provisioned there is no constant cost. The cost is directly proportional to the number of times the particular function is called and the amount of resources it consumes. This model is useful as for a startup with limited resources or for a big company who want to deploy the applications quickly.

Here is a nice video form the founder of aCloudGuru on how they built their entire business on Serverless Architecture. Here is a nice article/tutorial from AWS on using Lambda function to shrink the image uploaded to S3 bucket.

In the above workflow as soon as a image is uploaded to S3 Source Bucket, it will fire an event and call the Lambda function. The function will shrink the image and then put it in the S3 Target Bucket. In this scenario, there is no sustained cost associated as we pay based on the number of times the function is called and the amount of resources it consumes.

For the last few days I was intrigued by the Serverless architecture and had been trying the use cases mentioned in AWS Lambda documentation here. It was fun, but not pretty straight forward or as simple as uploading a function. The function has to be authored, packaged, uploaded, set the permissions for the event/resources, test the Lambda and finally integrate with the API Gatway. oooops. It's not an impossible task, but definitely tedious. Good that I haven't mentioned the debugging of the Lambda functions, which is really a pain.

For the rescue is the Serverless framework which makes working with the Lambda functions easy. Setting up the Serverless frameowork on Ubuntu was a breeze. Creating a HelloWorld Lambda function in AWS with all the required dependencies was even easier.

Note that the Serverless framework supports other platforms besides AWS. But, in the this blog I will provide some screen shots with a brief write-up on the same. Here, I go with the assumption that Serverless framework and all the dependencies like NodeJS, integration with the AWS Security Credentials has been done.

So, here are the steps.

Step1: Create the skeleton code and the Serverless Lambda configuration file.

Step 2: Deploy the code to AWS using the CloudFormation.

Once the deployment has been complete, the AWS resources will be created as shown below in the CloudFormation console.

As seen in the below screen, the Lambda function also gets deployed.

Step 3 : Now is the time to invoke the Lambda function again using the Serverless framework as shown below.

In the Lambda management console, we can observe that the Lambda function was called once.

Step 4 : Undeploy the Lambda function using Serverless framework.

The CloudFormation stack should be removed and all the associated resources should be deleted.

These completes the steps for a simple demo of the Serverless framework. A couple of things to note that we simply uploaded a Python function to the AWS Cloud and never created a server and so the Serverless Architecture. Here a few more resources to get started with Serverless Architecture.

On a side note I was really surprised that the Serverless framework has been started by CocaCola and then at some point of time they decided to open source in Github. All the time it was companies like Facebook, Google, Twitter, LinkedIn, Netflix, Microsoft :) opening the internal software to the public, but this was the first time I saw something from CocaCola.

Maybe I should drink more CocaCola and promote them to publish such cool frameworks.

No comments:

Post a Comment