Customer Review System

August 28, 2022 6:00 PM PDT

---

This document summarizes the interview focused on designing a customer review system for food delivery services. The discussion covered functional and scaling requirements, system design choices, and considerations for data management and user interaction. The interview aimed to assess the candidate's ability to design a scalable and efficient system while addressing potential challenges.

Interview Details

• Topic: Customer Review System
• Level: L5 (Senior)
• Duration: 45 minutes

Requirements

Functional Requirements

• Customer review system for food delivery.
• Display average ratings for each restaurant.
• Accept text reviews and ratings (images may be added later).
• Allow users to provide and retrieve reviews without approval.

Scaling Requirements

• Daily Active Users (DAU): 10 million (with a total of 100 million users).
• Restaurants: 1 million.
• Food items: 50 million.
• Total reviews: 1 billion.
• Queries per second (QPS): 100 (10 QPS for writes).
• Estimated data volume: 40 GB of reviews per day.

System Design

High-Level Architecture

• Separate read and write services.
• Utilize NoSQL for eventual consistency and scalability.
• Recommended database: Cassandra for its ability to scale and manage partitions.

Database Schema

1. Comments_by_item

   • itemId (UUID)
   • commentId (timeUUID)
   • User_id
   • Comment (text)
   • Rating (0-10)
   • Picture1_location
   • creation_time

2. Rating_by_item

   • Item_id
   • Total_score
   • Total_count
   • Average score computed periodically (every 10-30 minutes).
   • Unsaved information can be stored in Redis.

Pagination and Popularity Features

• Implement pagination for reviews with limit and offset.
• Support for upvote/downvote to determine helpfulness of comments.
• Consider automatic detection of popular comments based on content analysis.

Handling Popular Reviews

• Use a flag for promoted comments.
• Backend system to scan comments for upvotes.
• Maintain a threshold for upvotes in the item table.

External APIs

• Discussion on the need for API design and how to handle new requirements.

Feedback and Discussion Points

Interviewer Feedback

• Soft skills were satisfactory; multiple options were considered.
• Suggested improvements on QPS understanding and API design.
• Emphasized the importance of time management during the interview.

Interviewee Reflections

• Time control was not ideal; key points were missed due to time constraints.
• API design discussions took longer than expected.
• Suggested that understanding the interviewer's focus could have streamlined the process.

Audience Insights

• Discussed the importance of including user and restaurant information in data modeling.
• Addressed the risks of denormalization and the need for separate tables for comments.
• Highlighted the need for a ranking algorithm for comments based on popularity and relevance.

SQL vs NoSQL Considerations

• Discussed trade-offs between SQL and NoSQL databases, focusing on consistency, scalability, and access patterns.
• Noted that NoSQL offers flexible schemas, which can be beneficial for evolving requirements.

Conclusion

The interview provided valuable insights into the design and implementation of a customer review system. The discussions highlighted the importance of scalability, data management, and user experience in building a robust system. Further exploration of API design and ranking algorithms could enhance the overall architecture.