N-Tier Architecture

Overview

• The number of tiers is based on the number of layers (presentation, application, data, ...), not the number of physical computers used
• Multiple layers can exist on the same computer

Program logic

• UI code: Displaying UI elements and handling UI events
• Database code: Fetching and storing data in a database
• Business logic: Processing the fetched data based on UI commands

Single-tier

• The database and program logic runs on a single machine (standalone, or accessed via a dumb terminal)

Two-tier

(client-server)

• The database runs on a central server and most or all of the program logic runs on a client computer (a workstation)

• Pros
  

• Puts less of a load on the server

• Cons
  

• Requires more-powerful client computers (fat clients)
  
• Difficult to update the business logic over a wide range of client computers
  
• All clients are susceptible to lack-of-service if the server is down

Three-tier

(thin client)

• Database runs on a server (data layer)
  
• Business and database logic runs on a middle-tier server (application layer) (i.e. application server, middleware)
• User interface runs on the client (presentation layer)

• Thin clients consist almost entirely of the user interface
• If a web browser is used for the user interface, a wide range of computers can be used as a thin client
• The system is easy to maintain as long as the interface between the presentation and application layers doesn't change

• Low bandwidth requirement between clients and middle-tier servers
• High bandwidth requirement between database and middle-tier servers

• Pros
  

• Lightweight client computers (easy to maintain)
• Low bandwidth requirements (better scalability)

• Forced separation of UI and business logic
• Business logic is confined to a single machine

• Redundant application and data servers
• Data replication and synchronization
• Load balancing