Design Google Calendar

June 11, 2023 7:00 PM PDT

---

This document outlines the system design considerations for a Google Calendar-like application, focusing on requirements, database design, API specifications, and optimization strategies.

Requirements

• Total users: 500M
• Daily Active Users (DAU): 100M
• Queries Per Second (QPS):
  • Write: 1000 QPS
  • Read: 10,000 QPS

API Design

Booking Event

• Request Parameters:
  • Sender
  • Invitee
  • Title
  • Time
  • Location
  • Text
• Endpoint: GET /calendar-event

Considerations

• Handling anonymous users
• Supporting multiple invitees (email or ID)
• Read/write throughput for a single machine
• Updating meeting times

Database Design

Relational Database Schema

• Events Table:

  • eventID
  • sender
  • invitee (string)
  • title
  • time
  • location
  • text

• Normalization:

  • Factor out event detail table to avoid repeated fields.

Data Volume Calculation

• Estimated data volume:
  • 1.5TB per year
  • 15TB over 10 years

Primary Key Definition

• Primary Key: sender + eventID
• Index: dayID
• Composite Index:
  • participant + dayID

Query Optimization

• Range query for meetings:

  SELECT * FROM meetings WHERE time > from_ts AND time < to_ts
  

• Use of indices to optimize queries, particularly for range queries.

Handling Multiple Invitees

• Considerations for multiple rows for invitees.
• Handling changes in event times may require updates to multiple rows.

Email Notification System

• Implement an email notification system for users outside the system.
• Use a meeting processor for asynchronous changes to events.

Database Technology Considerations

• Relational vs Non-relational databases:
  • High throughput and volume may necessitate sharding by time range.
  • Non-relational databases may be recommended for handling large event tables.

Conflict Resolution

• Addressing conflicts when multiple users have meetings at the same time.
• Gathering requirements for supporting meeting rooms.

Additional Considerations

• Sharding strategies using user ID or invitee ID as partition keys.
• Microservices architecture for better scalability and maintenance.
• Strategies for syncing data to clients and indexing the database effectively.