Architectural Design(part 2)

Control Model

A control model deals with control flow between the subsystems. In order to make a system work effectively, its constituent sub systems must be controlled to ensure that the services are delivered in a precise manner. There are two types of control models. They are:

A)Centralized Control Model

In this model, one of the subsystems is designated as system controller with responsibility for managing the execution of other subsystems. Depending on whether the controller system executes sequentially or in parallel, the centralized controlled models are of two types:

a)Call- return Model: This is a top- down sub-routine model where the control passes from a higher level subroutine in a hierarchy to the lower level routine. This model is only applicable to sequential system. In this model, currently executing subroutine has the responsibility for control. It can either call other subroutines or returns control to its parent.(Sommerville SE_book)

b)Manager Model: This is applicable for concurrent systems. One system component is designed as a system manager that manages the starting, stopping, and coordination of other system processes. A process is a subsystem that can execute in parallel with other processes. It is also applied in sequential systems where a management routine calls a particular subsystem depending on the values of same state variables through case statement. It is useful in soft-real time systems where time is not a tight constraint. It requires very powerful algorithm to control concurrent processes.

B)Event -Driven Model

These models are driven by externally generated events. There are two types of such models which are:

a)Broadcast Model: These are appropriate for integrating subsystems distributed across different computers on a network. In this model, an event is broadcast to all the subsystems. Any subsystems which can handle that event respond to it. The event and message handler maintain a register of subsystems and the events of interest to them. The event handler detects the event to those subsystems that have registered an interest in the event. A subsystem can send a message to another subsystem.

b)Interrupt-Driven Model: These are exclusively used in real time system with stringent timing requirement. The external events are detected by an interrupt handler. In an interrupt driven model, the interrupt types are identified with their respective handlers. Each interrupt is associated with the memory location
where handlers address is stored.

Domain Specific Architecture

The above architecture models are general models. They can be applied to many classes of application. As well as those general models, architectural models that are specific to particular application domain are called domain specific architecture.

There are two types of domain-specific architecture models:

1. General Models: These are abstractions from a number of real systems. They encapsulate principal characteristics of these real-time systems. These may be reused directly in a design. Eg: compiler model.
2. Reference Model:These are more abstract and describe a larger class of systems. These include all the features that system might incorporate. These are used to communication domain concepts and compare or evaluate possible architectures. It may be used as a basis for system implementation. Eg OSI model.

Difference between Generic and Reference Model

• Generic model may reuse directly in a design.
• Reference model are normally used to communicate domain concept and compare possible architectures.
• Generic models are usually derived “bottom-up” from existing system.
• Reference modes are derived from “top-down”.
• Generic models are abstract system representations.
• Reference model do not necessarily reflect the actual architecture of existing system in the domain.

Reference: