Evolution of Software Architectures

by K. Yue
August, 2002, last modified: july 1, 2004
copyright 2002-4

1. Introduction: Why semantic Web?

• Intelligent processing of information will be possible.
• Low noise to signal ratio.
  • More efficient transmission.
  • Easier to process.
• More personalized information.
• Explosive growth of information communication.
• Internet data will increasingly be processed by machines, and not viewed by human beings.
• There will be much richer way for human computer interaction (HCI).
• Huge number of new applications will appear.

2. Evolution of Software Architectures

• Software architecture has changed during the last 40 years.
• A very coarse evolution will be shown.

2.1 Phase 1: Mainframe-based

• Highly centralized.
• Very thin clients: dumb terminals.
• Largely disconnected to outside of the organization.

2.2 Phase 2: Traditional 2-tiered Client and Server

• Fat clients: PC.
• Proprietary protocols.
• Many dedicated servers.
• Access largely within the organization.
• Interoperability not yet a large issue.

Figure 1 Basic organization of traditional 2-tiered client and server system.

2.3 Phase 3: Web-Based N-tiered Client-Server

• Open Internet protocols.
• Middle layer for definition of business logic: better software development and management.
• Serving clients outside of the organization: Internet.
• The browser was the "universal client".
• Thinner clients: no need of complicated proprietary client software.

Figure 2 Web-based N-tiered client and server system

2.4 Phase 4: Web-Centric Applications

• Potentially thin clients.
• Clients and servers can be blurred.
• Services in and out of the organization.
• Data/services from many servers, inside and outside of the organization.
• No single organization can provide 'everything.'
• Much more personalized software.
• XML is a cornerstone technology of Web-Centric architectures.
• Web services become popular.
• Semantic web connects applications.
• Peer-to-peer technology.

Figure 3 Web-centric system