Computer System Design Intro

Computer System Design Introduction

• Introduces vocabulary and concepts for designing computer systems
• Introduces a "systems perspective" considering global rather than particular issues

What Computer System Design is about

• Focuses on both computations (software) and physical constructs (hardware)
• Covers areas like theory of computation, AI, computer architecture, software engineering, compilers, networks, and security
• Assumes students have introductory courses on software and hardware completed
• Begins a broad study of computer systems that supports the entire range of systems specialties

Why Computer System Design

• Required for interesting computer applications
• Needed for fault tolerance, coordination of concurrent activities, geographically separated but linked data, vast information storage, protection, and interactions with many people
• Designers must look beyond software and hardware, viewing the entire computer system as a whole

System Design Principles

• Applications should be developed with these requirements in mind with a view to the overall computer system including Software and Hardware
• Design principles are guidelines to avoid making mistakes

Systems

• Defined as collections of elements/components organized for a common purpose
• Classification based on mathematical complexity: linear and nonlinear

Linear Systems

• Output change is proportional to input change
• Satisfies the superposition principle, including additivity and homogeneity
• Additivity examples: F(x1) = Y1, F(x2) = Y2, F(x1 + x2) = Y1 + Y2
• Homogeneity examples: F(x) = y, F(ax) = ay ("a" is a scaler)
• Generally simple and do not change over time

Nonlinear Systems

• Output change is not proportional to input change
• Of interest to engineers, biologists, physicists, and mathematicians
• Commonly nonlinear in nature
• Dynamical examples exhibit chaotic, unpredictable, or counterintuitive behavior

Systems And Complexity

• Encompasses common problems, systems, components, interfaces, environments and complexity

Common Problems of Systems

• Can be divided into emergent properties, propagation of effects, incommensurate scaling, and trade-offs

Emergent Properties

• Properties that are not evident in individual components alone; also called "surprises"
• e.g., Committee behavior, electric power distribution failures

Propagation of Effects

• Small disruptions or local changes that affect the entire system
• e.g., Domino chain falling, tree falling on a power line in Oregon causing lights to go out in New Mexico
• Systems must be designed to limit the impact of failures

Incommensurate Scaling

• Not all parts of a system follow similar scaling rules
• Different system parts exhibit different orders of growth
• Scaling a mouse to elephant size will cause it to collapse under its weight
• Supertankers take 12 miles to stop, going beyond the horizon view from the bridge

Trade-Offs

• Limited resources in the universe (system) leads to design challenges
• Aim is to maximize usefulness, avoid waste and allocate appropriately
• Pushing hardware circuit clock rates increases power consumption, heat, and timing errors

Systems, Components, Interfaces, and Environments

• Systems: Interconnected components with an expected behavior at the interface with its environment
• Components: Individual parts that make system
• Interfaces: Provide interactions between the system and environment
• Environment: Everything outside the system boundary, affecting it.
• PC interfaces: display, keyboard, network connection, power cord, atmospheric pressure, ambient temperature, humidity, electromagnetic noise

Views of a Jet Aircraft as a System

• Engineer view: System includes body, wings, control surfaces, and engines, environment is the atmosphere and earth while interfaces include gravity, thrust, and air drag
• Company/Passenger view: System includes seats, flight attendants, and air conditioning; the environment is the passengers and the interfaces include soft seats, meals, and air quality

Complexity

• Defined as "difficult to understand"
• Signs that a system is complex is having a large number of components, large number of interconnections, many irregularities, long descriptions or a team of designers, implementers, or maintainers

Sources of Complexity

• Two primary sources include cascading/interacting requirements and maintaining high utilization

Cascading and Interacting Requirements

• The list of requirements for a system becomes a complexity source
• The accumulation of many requirements adds their complexities and the complexities from their interactions

Maintaining High Utilization

• Starts with a desire for high performance or efficiency using limited resources

Utilizing Railroad as an example

• A single-track railroad line passes through a long canyon
• Improving single-track usage means trains run both ways
• To achieve it installing a switch and a short side track helps one way traffic flow at one time

Reducing Complexity

• Modularity, abstraction, layering, and hierarchy are common techniques

Modularity

• Design system as interacting subsystems or modules
• Debugging a large program takes an amount of time, and dividing the program into modules debug time is reduced
• Modularity allows for easy replacement of an inferior module
• Modular programs also allow incremental system improvements
• It is easier to change a module than to change the modularity

Abstraction

• Separate interface from internals and specifications from implementation
• Often paired with modularity
• Functional modularity is a term used for modularity with abstraction

Abstraction Examples

• DVD Player: If one needed to know the details of a television set to choose a compatible DVD Player
• Delivery Company: Delivers without needing to know vehicle types or routes

Abstraction Examples In the Computer World

• Flip-flops: We intentionally hide all the details
• Design: Hides details of complex data structures and detail
• Users interact with simple button-pushing interfaces and do not manage processor, memory, and comms

Layering

• Reduces module interconnections
• Systems use good abstractions to minimize interconnections

Hierarchy

• Technique which reduces interconnections in a different, specialized way
• Assemble modules into stable, self-contained subsystem with a well-defined interface
• Group subsystems become larger subsystems

Hierarchy examples

• Corporations have managers supervise 5~10 employees
• And OOP - Object Oriented Programming

Connection thru Names:

• Modularity, abstraction, layering and hierarchy divide things into modules
• A way of connecting these modules

Digital Systems primary Connection:

• One module names another module

Computer Systems, same but different

• Computer Systems Designer can investigate techniques from other fields that may also apply to computer systems
• The complexity of a computer system is not limited by physical laws
• The rate of change of computer system technology is unprecedented

Computer Systems have no Nearby Bounds on Composition

• Computer systems are mostly digital, controlled by software
• Limits on complexity are relaxed compared to systems limited by physical laws
• Analog systems, each component contributes noise whereas digital are noise free, so complexity can grow freely

Reducing Complexity via Iteration

• Major help from Modularity, Abstraction Layering and Hierarchy that are insufficient to control resulting complexity
• iteration developed to cope with more complex computer systems

Iteration

• Starts with simple working subset of requirements (prototype) -> evolve via small steps to encompass more requirements
• Release indicators like "release 4.301"

Iteration Design Principles

• Requires considerable foresight and involves elements found during iteration such as taking small steps and not rushing
• Smaller steps help discover design defects and bad ideas
• Steps must also be well-planned, avoiding the temptation to rush implementation

Iteration Design

• Feedback should be planned
• Feedbacks from multiple people can refine a design
• Success starts with failures. and studying them
• It is key to focus on failures that must be studied, avoided and planned for