Atelier Product Q&A API Service

• Redesign and optimization of a monolithic backend system to a set of microservices for an e-commerce web application.
• Responsible for the Product Q&A microservice that supports the frontend Questions & Answers module that allows users to view and search questions, ask questions, answer questions, and provide feedback on questions and answers for a given product.

Tech Stack

API Endpoints

Questions

GET /qa/questions retrieves a list of (non-reported) questions for a given product
POST /qa/questions adds a question for a given product
PUT /qa/questions/:question_id/helpful updates a question to show it was found helpful
PUT /qa/questions/:question_id/report updates a question to show it was reported

Answers

GET /qa/questions/:question_id/answers retrieves a list of (non-reported) answers for a given question
POST /qa/questions/:question_id/answers adds an answer for a given question
PUT /qa/answers/:answer_id/helpful updates an answer to show it was found helpful
PUT /qa/answers/:answer_id/report updates an answer to show it was reported

Performance Summary & Stress Testing

Metric	Goal	Achieved
RPS	1000	1000 for questions, 1500 for answers
Latency	<1s	26ms for questions, 12ms for answers
Error rate	<1%	<1% for both questions and answers

Deployed and stress tested with loader.io, with bias on records within last 10% of dataset

GET questions /qa/questions

GET answers /qa/questions/:question_id/answers

Optimizations

Server optimizations

• Horizontal scaling with 1 load balancer and 2 server instances (total of 3 AWS EC2 T2 micro instances)
• Load balancer configurations: keepalive_requests and Nginx caching

Other optimizations

• Database optimizations with indices (query time <50ms)
• Server optimizations with Redis caching