Distributed-Reservation-System

Distributed-Reservation-System

High level features

1. User browses through Rooms available for a given date range
2. User reserves a type of Room in a particular Hotel
3. On check-in, hotel manager assigns a Room of that type to the user

Performance Considerations

When do WRITE operations happen ?

1. User reserves a Room
2. User cancels a resevation
3. New Hotel or room is added

When do READ operations happen ?

1. Browsing through hotel catalog
2. Browsing through hotel features

• Our system is read-heavy -- Significantly higher amount of READ than WRITE.

API Requirement

Reservation

• GET /api/v1/reservations
• GET /api/v1/reservations/123
• POST /api/v1/reservations
• DELETE /api/v1/reservations/123

Data Model

• Let's go with a relational database i.e PostgreSQL
• Why:

1. ACID properties, transactional guarantees
2. Easier Locking mechanisms
3. Data can be easily sharded for scalability eg based on the hotelIDs
4. Easier to model hotel and resevation data
5. More READs than WRITEs
6. Mostly CRUD Operations

Architecture diagram