COMP2120 Week 4 - Architecture

Table of Contents

Twin Peaks View

As more is known about a project, the higher the peaks grow. They progressively develop more detailed requirements and architectural specifications concurrently.

Essentially the same as 'design'
Architectural design -> Designing the overall structure of the system
- Big picture
- Identifies main structural components and the relationships between them

A creative process, where design outcomes differ depending on the type of system being developed
Common decisions:
- Is there a generic application or pattern that can act as a template?
- How will the system be distributed or run?
- How will components be split into sub-components?
- What architecture satisfies the non-functional requirements?

Performance -> Localised critical operations and minimised communications. Not very distributed
- Sommerville Chapter 17
Security -> Layered architecture with critical assets at innermost layers
- Sommerville Chapter 13
Safety -> Localised safety-critical features all in one component.
- Sommerville Chapter 12
Availability -> Redundant components and mechanisms for fault tolerance
- Sommerville Chapter 11
Maintainability -> Fine-grain, replaceable components.

Note: Performance needs large components, but maintainability needs small components

Best "logical" architecture
If in doubt, start here
Set of layers which provide services and can support incremental development of different layers independently of other layers
- Rules can be defined
- Typically "higher" layers depend on "lower" layers
  - Strict layering -> A layer only depends on the layer immediately below it

Name	Description
Description	Organised into layers with related functionality. Layers provide services to level above it sothe lowest-level layers represent core services
When is it used	Good generic architecture. Good where development is spread across several teams + requirement for multi-level security
Advantages	Whole layers can be replaced as long as the interface is maintained. Redundant facilities can be provided in each layer to increase dependability
Disadvantages	Clean separation of layers is difficult. Performance can be a problem due to multi-level inerpretations of a service request (processing at each layer)

Name	Description
Description	Separates presentation and interaction from system data. Three logical components. Thew Model component manages system data and operations on data. The View component defines and manages how the data is presented. The Controller component manages user interaction and passes them to both the View and Model
When is it used	Best used when multiple possible ways to view and interact with data are possible.
Advantages	Allows data to change independently of its representation and vice versa. Presentation of the same type of data in different ways with changes made in one shown in all of them
Disadvantages	Additional code and code complexity required. Application can turn complex and become a mess

Name	Description
Description	All data managed in a central repository accessible to all system components. Do not interact directly, only via repository
When is it used	Use if large volumes of information is generated that must be stored for a long time. Can also be used in data-driven systems
Advantages	Components can be independent. Changes made by one componenet can be propagated to all components. All data can be managed consistently
Disadvantages	Repository is a single point of failure. Inefficiencies in organising communication through repository may exist. Discribution of repository across multiple devices is difficult

Distributed system model
Can be stand-alone servers which provide specific services, with a set of clients who use those services

Name	Description
Description	Functionality organised into services, delivered from a seperate server.
When is it used	Best suited for use in a shared database where it will be accessed from a range of locations. Server replication allows for upscaling of services based on a variable load. Anticipates microservices
Advantages	Principle advantage is distributed servers across a network where it doesn't need to be implemented by all individual services
Disadvantages	Each service is a single point of failure. Performance depends on the network as well as the system

Functional transformations process inputs to become outputs
Useful when transformations are sequential (batch sequential model) and is extensively used in data processing systems
Not realy suitable for interactive systems

Name	Description
Description	Processing is organised so that each processing component is discrete and carries out one type of data transformation. Data flows (like a pipe) from one componenet to another
When is it used	Data processing applications where inputs are processed in separate stages to generate related outputs
Advantages	Easy to understand and supports transformation reuse. Can be implemented as either a sequential or concurrent system
Disadvantages	Format for data transfer needs to be pre-agreed on. Transformations need to parse its input and unparse output to agreed form, increasing overhead