A team is developing a feature that needs to recognize Celebrities. By using the App, clients can upload photos and search celebrities among the photos by clicking a button. Or they can upload a bunch of photos and search the times that a given celebrity has appeared. The team wants to run the App in AWS at a lower cost. Which option is the most efficient one to implement while still ensuring availability and stability?

Answer options:

A.Create the frontend and backend in a T2 medium EC2 instance to use its burstable capability. Call Rekognition API “RecognizeCelebrities” to fetch the information in a JSON format. Process the JSON result in the backend service and return the result to the frontend UI.
B.Develop the App in a serverless lambda to use Rekognition API “SearchFaces” to search a Celebrity. The input image can be base64-encoded bytes or an S3 object. After the API has been returned, present the result to clients.
C.Use the AWS Rekognition service. Implement the App in a lambda to call Rekognition API “RecognizeCelebrities” to fetch the information required in a JSON format. Process the information in Lambda and return the result to end-users. Use S3 for clients to upload photos.
D.Implement the App in an M4 large EC2 instance with Autoscaling and Elastic Load Balancer. Build the application via a Cloudformation template. Use the App to call Rekognition API “SearchFaces” to get the information. Process the JSON result in the backend service and return the result to the frontend with a Cloudfront CDN.
 

Correct Answer – C
 
As the question asks for a lower cost while still ensuring availability and stability. Lambda should be considered first. Lambda can autoscale without manual operations. More importantly, it is very efficient to develop a web app without considering a lot about OS, patching, etc.
Amazon Rekognition can recognize thousands of celebrities in a wide range of categories, such as entertainment and media, sports, business, and politics. With Amazon Rekognition, you can recognize celebrities in images and stored videos. You can also get additional information for recognized celebrities.
Option A is incorrect: Because EC2 does not cost-efficient and brings in extra cost if the load is not stable. Lambda should be considered instead.
Option B is incorrect: Because Rekognition API “RecognizeCelebrities” should be used in this case. For each celebrity recognized, “RecognizeCelebrities” returns a Celebrity object. The Celebrity object contains the celebrity name, ID, URL links to additional information, match confidence, and a ComparedFace object that you can use to locate the celebrity`s face on the image. “SearchFaces” is incorrect as it is not used to search Celebrities.
Option C is CORRECT: Because Lambda together with Rekognition can meet the need for low cost, availability, and stability. Refer to https://docs.aws.amazon.com/rekognition/latest/dg/celebrities.html for this feature.
Option D is incorrect: Because firstly, it has used EC2. Even if Autoscaling and ELB are used, it is not as efficient as Lambda. Secondly, API “SearchFaces” should not be used for the function of Celebrities Recognition.