Table of Contents
To revamp the existing Questions and Answers service on the e-commerce platform, enhancing its capabilities, scalability, and efficiency.
In this project, I replaced the existing API with a backend system that can support the project's full data set and scale to meet the demands of production traffic. I ensured that the backend system was optimized for scalability to guarantee reliable performance, even during high-traffic periods. It was essential to ensure that our system not only supported the current user base but was also future-proofed to anticipate and accommodate significant user growth. The e-commerce platform is now better equipped to handle user queries efficiently, even as the platform's user base expands exponentially.
Scaled to 1000 requests per second for random sampling and bias selection for records within the last 10% of the dataset, which contains 100,000 records.
The average response time is less than 3ms, and the error rate is 0%.
Optimizations: PostgreSQL indexing, load balancer with 4 AWS EC2 servers running
qa/questions?product_id=%{:900000-1000000} || 1000 requests per second || 1ms 0% error rate
qa/questions/%{:900000-1000000}/answers || 1000 requests per second || 3ms 0% error rate
Scaled to 1000 requests per second for random sampling and bias selection for records within the last 10% of the dataset, which contains 100,000 records.
The average response time is less than 6ms, and the error rate is 0%.
Optimizations: PostgreSQL indexing, load balancer with 4 AWS EC2 servers running
qa/questions || 1000 requests per second || 6ms 0% error rate
qa/questions/%{*:1-100000}/answers || 1000 requests per second || 6ms 0% error rate
Scaled to 10k requests per second for records within the newest 100 product_id in the dataset.
The average response time is less than 3ms, and the error rate is 0%.
Optimizations: PostgreSQL indexing, load balancer with 4 AWS EC2 servers running, NGINX Content Caching.
qa/questions?product_id=%{*:999900-1000000}
