Serverless vs EKS/EC2

Question

Dear Team,

We are building a service which will be exposed as Rest API through API Gateway.  The service will perform 3 things.

1. Validate the inputs
2. Upload a document to S3(rather move the document-approx 40kb in size from Staging S3 location to Target S3)
3. Insert the metadata to Dynamo DB

We expect this service to be called approx 2 lakh times within 8-10 hours on daily basis through AWS Batch.  Afterwards, it may not be used through out the day.  We were planning to have a lambda function for each of the above steps which will be called from API Gateway.  If required, we may use Step functions to orchestrate above 3 steps.

Would Lambda functions be the appropriate way to process these steps in the service or should we use EC2/EKS and ASG to scale in and Scale out.

I assume the service will respond back within second but bearing Lambda concurrency limits in mind and any other limits we may face, let me know the best way to process the batch.

Regards,
Dhaval Mehta

Accepted Answer

Yes, using Lambda here, see documentation - https://docs.aws.amazon.com/lambda/latest/dg/services-s3-batch.html

There is also a blog post on batch processing - https://aws.amazon.com/blogs/compute/creating-aws-serverless-batch-processing-architectures/

And some best practices - https://docs.aws.amazon.com/lambda/latest/dg/best-practices.html

Answer

To me this is a perfect use case for Lambda. +1

Answer

If you build your Lambda function with containers, then you can consider running the containers in EKS in future should your use case change.

Serverless vs EKS/EC2

Relevant content