Design a Calendar System

Scenario

Design the backend architecture for an enterprise-scale calendar service (similar to Google Calendar or Outlook Calendar) that integrates into a professional network. Users can schedule individual or recurring meetings, invite colleagues, update RSVP statuses, and receive scheduled reminders.

Requirements

Functional Requirements

1. Event Management: Users can create, update, and delete single and recurring events (e.g., daily, weekly, custom iCalendar RRULE patterns).
2. Calendar View: Users can view all events on their calendar within a specified time range (e.g., day, week, month view).
3. Invitations & RSVP: Organizers can invite other users. Invitees can view invitations and respond with Accepted, Declined, or Tentative.
4. Reminders: Users receive push or email notifications configured for X minutes prior to an event.

Non-Functional Requirements

1. Scale: 100 million daily active users (DAU).
2. Performance: Querying a 1-month calendar view should return in under 100 ms.
3. Availability & Consistency: High availability for reads; strong consistency for attendee invitation state and event modifications to avoid phantom meetings.
4. Timezone Support: Correct representation of events across different time zones and daylight saving transitions.

Scale & Traffic Estimations

• DAU: 100 million.
• Writes: Average 1.5 event creations or modifications per active user per day (~1,800 writes/sec average, peak ~5,000 writes/sec).
• Reads: Average 8 calendar views per active user per day (~9,500 reads/sec average, peak ~25,000 reads/sec).
• Reminders: Hundreds of millions of notifications evaluated and sent daily.

What Your Design Should Cover

• API Design: Core endpoints for creating events, fetching calendar windows, and responding to invites.
• Data Model: Storage schema separating master recurring events, recurring event exceptions, and attendee status.
• Recurrence Architecture: Strategy for expanding recurring events (pre-computed/materialized vs. dynamic expansion on query).
• Notification Pipeline: Distributed reminder engine capable of triggering accurate alerts at high throughput.
• Trade-offs: Storage cost vs. query latency when indexing future events.