Future Directions in Middleware Research and Technology

1
Future Directions in MiddlewareResearch and
Technology

• Fabio Kon
• kon_at_ime.usp.br
• Department of Computer Science
• University of São Paulo, Brazil
• http//www.ime.usp.br/kon

2
Main Issues

• Component-Based Software
• Automatic Configuration
• Adaptive Systems
• Separation of Concerns / Reflection
• Ubiquitous Computing
• Social Aspects

3
Component-Based Software

• Development we need tools for
• writing components
• assembling components
• debugging and testing
• Deployment
• standard environments
• INTEROPERABILITY
• Systems
• Education

But, its still our only alternative!
4
Automatic Configuration

• Past many users per computer
• Present few computers per user
• Future hundreds (thousands ?) computing devices
  per user
• Solutions
• 1. Static configuration from manufacturer
• 2. Automatic Configuration

5
Recent Advances in Automatic Configuration

• 2K some ideas, a nice prototype, many problems.
• Debian Linux limited scope, very robust and
  extremely useful implementation.
• Major problem how to specify what a component
  needs (prerequisites).
• Prerequisite specification language
• Prerequisite generator
• (in fact, we need meta-information about
  components, systems, and environments)

6
Future Work in Automatic Configuration

• Study prerequisite specification in Enterprise
  Java Beans, CORBA Component Model, .NET.
• Investigate
• how to generate them automatically
• how to use them for automatic configuration.

7
Adaptive Systems

• One size does not fit all.
• Environments are becoming more and more dynamic.
• So, we need
• 1. generic model/framework for building adaptive
  systems
• 2. compose repertoire of adaptations when
  designing new systems.

8
Separation of Concerns

• Functional Aspects vs. Non-Functional Aspects
• Traditional approach (e.g., OO) is not enough
• Reflection seems to be the most effective and
  flexible way of doing it.
• Aspect-Oriented Programming (AOP)
• a promising idea
• but still very limited

9
Ubiquitous Computing

• It will happen
• most solutions are still ad hoc
• Gaia OS

10
But

• Why are we doing this?

• For whom we are doing this?

• Bill Gates and co.?
• Multi-billion companies?
• The 10 of the people who have Internet access?

• What about the other 90?

• Open Software!
• Free Software!