Chapter 14 System Design Prepared By: Ashwin Raiyani.

Chapter 14 System Design Prepared By: Ashwin Raiyani

Topic Covered Overview of system design Estimating performance
Making a Reuse plan Breaking system into sub-system Identifying Concurrency Allocation of Sub Systems Management of Data Storage Handling Global Resources Choosing a Software Control strategy Handling Boundary Conditions Setting Trade-off Priorities Common Architectural Style

System Design Till now, focus was on WHAT needs to be done, independent of HOW it is done. During System design, decide how the problem will be solved. Need to apply high level strategy – System Architecture – for solving the problem and building a solutions. During System design, developer decide the overall structure and style. System architecture provides the organization of the system into subsystem.

System Design To construct system architecture, following decision must make: Estimating performance Making a Reuse plan Breaking system into sub-system Identifying Concurrency Allocation of Sub Systems Management of Data Storage Handling Global Resources Choosing a Software Control strategy Handling Boundary Conditions Setting Trade-off Priorities Common Architectural Style

Estimating Performance
System Design Estimating Performance You should prepare rough performance estimate. Purpose of this task to determine if the system is feasible. You will have to make simplifying assumptions.(i.e. assume factors) Don’t worry about detail – just approximate, estimate and guess.

System Design For ATM exa. Consider following case
Bank has 40 branches and no. of terminals. On a busy day half the terminal are busy at once. Suppose each customer takes one min to perform a session and most transaction involve a single transaction. i.e. 40 transaction a time. You can perform similar estimate for data storage. Count the no. of customer, estimate amount of data for each one and multiply. i.e. 40 *20 =800.

Making a Reuse plan System Design Two different aspect of reuse
Using existing thing Creating reusable new things. Much easier to reuse existing things than to design new things. Most developer reuse existing thing and only a small fraction of developers create new things. Creating reusable new things is not easy task. It require good experience.

System Design Reusable thing includes mainly
Libraries Frameworks Patterns Libraries: it is a collection of classes that are useful in many contexts. Classes must be organized, so users can find them. Online searching can help more. Classes must have accurate and thorough description to help users determine their relevance.

System Design Several qualities of good class libraries:
Coherence: organized well focus Completeness: complete behavior Consistency: consistent names and signature. Efficiency: provide alternative to implement Extensibility: able to define subclasses Genericity: should be parameterized class where appropriate.

System Design Framework: it is skeletal structure of a program that must be elaborated to build complete application. Framework consists of more than just the classes. It is more specific to individual application. Framework class libraries are typically application specific and not suitable for general use.

System Design Patterns: A pattern is proven solution to a general problem. There are patterns for analysis, architecture, design and implementation. A pattern comes with guideline on when to use it, as well as exchange on its use. There are many benefits. One advantage, carefully created and applied on past problems. A pattern is different from a framework.

System Design A pattern is typically a small number of classes and relationship. A Framework is much broader in scope and covers an entire application.

Breaking a system into SubSystems
System Design Breaking a system into SubSystems First steps in system design is to divide the system into pieces. Each major pieces of a system is called Sub System. Each subsystem is based on common functionality, common physical location and common hardware. A subsystem is not an object nor a function. But a group of classes, association operations, events and constraints that are interrelated.

System Design A Sub system always identified by its Service it provides. A Service is group of related functions that share common purpose. For. File system within an operating system is a subsystem. Each subsystem interface specifies the form of all interactions and information flow between them. Relationship between two subsystems can be client-server or peer-to-peer.

System Design Client-Server Relationship: Peer-to-Peer Relationship:
Client calls on the server, which performs some services and replies with a result. Client must know server interface, but server did not know its client’s interface. Because client initiates all interactions. Peer-to-Peer Relationship: A communication from one subsystem to another is not necessarily followed by an immediate response.

System Design Each subsystem must know each other’s interfaces. Therefore, these interactions are more complicated. So Decomposition of system into subsystem may be organized as a sequence of horizontal layers or vertical partitions.

System Design Layers: A layered system is ordered set of tiers.
Each built in terms of the ones below it and providing the implementation basis for the ones above it. Information is one way only – subsystem knows about the layers below it but has no information of layers above it. Client-Server relationship exists between upper layers(users of services) and lower layers (providers of services)

System Design In interactive graphics system, windows are made from screen operation, which implemented using pixel operation, which execute device I/O operation. Each layer is implemented in terms of classes and operation of lower layers. Layers architecture are in two forms: Open Architecture Close Architecture.

System Design Closed Architecture: Open Architecture:
Each layer built only in terms of immediate lower layer. Reduces dependencies between layered Allows changes easily. Open Architecture: A layer can use features of any lower layer to any depth. It reduces the need to redefine operation at each level. Changes to sub system can affect at higher level So, it is Less robust compare to close architecture

System Design Usually Problem statement specifies only the top and bottom layers: Top layers is desired system Bottom layers is available resources (Hardware, Operating system)

System Design Partitions:
Vertically divide a system into several independent or weakly coupled subsystem. For ex. Computer operating system includes a file system, process control, virtual memory management and device control. Sub system may have some information of each other but not deep.

System Design Diff. between layers and partitions
Layers vary in their level. While partitions only divide a system into pieces.( all have same level) Layers depend on each others(client Server relationship). While partitions are independent (peer to peer relationship)

Combining layers and partitions
System Design Combining layers and partitions Layers can be partitioned, and partitions can be layered. Most large system require a mixture of layers and partitions. (Refer fig 1) Once you have identified the top-level subsystems, you can show their informational flow. Many computations have the form of pipeline;

System Design Figure 1

System Design For ATM example: there are 3 major sub systems. ATM stations, Consortium Computer and Bank computes. Consortium computer communicate with all ATM and bank computers ( comm. Links) Architecture uses station code and bank code to differentiate phone lines and consortium computer.

System Design

Identifying Concurrency:
System Design Identifying Concurrency: Identify the objects that must be active concurrently and objects that have mutually exclusive activity. You can fold the latter objects into a single thread of control or task. Two steps: Identifying inherent concurrency Defining Concurrent Tasks.

Identifying inherent concurrency
System Design Identifying inherent concurrency State model guide to identifying concurrency. If two objects receive events at the same time without interacting then its called inherently concurrent. If event are unsynchronized, you can’t fold the objects into a single thread of control. Often problem statement specifies that distinct hardware units can implement the objects.

System Design For Example ATM: it was stated (in chapter 11) each machine should continue to operate locally in event of central system failure, then would have no choice but to include a CPU in each ATM machine with full control program.

Defining concurrent Tasks
System Design Defining concurrent Tasks By studying state diagram of individual objects and exchange of events among them. You can fold many objects onto a single thread of control.. A thread of control is path through a set of state diagram on which only a single object at a time is active.

System Design A thread remains within a state diagram until an object sends an event and passes to the receiver of event. A thread splits if the object sends an event and continues executing. Only single object at a time is active. ATM Ex. : if a central computer directly controls the ATM, we can combine ATM object with Bank transaction object as a single task.

Allocation of Subsystem
System Design Allocation of Subsystem Must allocate each concurrent subsystem to hardware unit. Either general purpose or specialized functional unit. Criteria would be Estimate hardware resource requirement Making hardware-software trade-off Allocating Tasks to processors Determining Physical connectivity.

Estimating hardware resource requirement
System Design Estimating hardware resource requirement System designer must estimate the required CPU processing power by computing steady-state load. I.e. Product of number of transaction per second and time required to process a transaction. You should increase estimate to allow for transient effects(i.e. failure cases) Steady-state and peak load are important.

System Design For. Ex. ATM machine just proive user interface and some local processing. So single CPU would work for each ATM. Consortium computer is essential – as it receives ATM request and sending them to appropriate bank computer. So it involves multiple CPUs. While Bank computers perform data processing and database application. Appropriate depend upon needed throughout.

Making Hardware-Software Trade-offs
System Design Making Hardware-Software Trade-offs You must decide which subsystem will be implemented in hardware and which in software. Two main reason: Cost – it is easier buy floating point chip than to implement floating point in software. Performance – more efficient hardware is available.

Allocating Tasks to processors
System Design Allocating Tasks to processors Several reasons for assigning task to processors Logistics : Some tasks are required at specific physical locations or to permit independent operations. For ex. At ATM, might communication network is down on that case ATM must have its own CPU and programming logic.

System Design Communication Limits Computation Limits
Sometime available comm. bandwidth between task & piece of Hardware is not sufficient then response is not getting on time or its taking more time to complete. Computation Limits Assign separates processors for computation process.

Determining Physical Connectivity
System Design Determining Physical Connectivity Connection topology: Choose the topology for connecting physical units. As Association in the class model indicating physical connection. For ATM communicate with Consortium. Also client-server relationship also correspond to physical connections.

System Design Repeated units
Topology has a regular pattern such as Linear Sequence , A martix, A tree, A Star If you have boosted performance for particular kinds of units or group of units, then you must specify their topology. Communications: Choose the form of connection channel & comm protocol.

System Design System designer may specify the exact interfaces among units for interaction mechanisms. For Ex. Interactions may be Asynchronous, Synchronous or blocking. Based on estimate bandwidth & latency of the comm channels and choose the correct kinds of comm channels.

Management of data storage
System Design Management of data storage There are several alternatives for data storage. You can use separately or in combinations. Data structure Files Database Based on cost, access time and reliability different kinds of data storage. For Ex. PC application may use memory data structure or files. An accounting system may use a database,

System Design Data storage – Files
Files are cheap, simple and permanent. Files implementation vary for different computer systems. For ex. Implementation for sequential files are mostly standard, but commands and storage formats for random-access files and indexed files vary

System Design Kinds of data that belongs in files Data with high volume and low information density (historical records) Quantities of data with simple structure. Data that are accessed sequentially. Data that can be fully read into memory

Data storage – Database
Various type of DBMS available Relational DB OO DB. DBMS cache frequently accessed data in memory. DB makes application easier to port different hardware and OS. Disadvantage – complex interface

System Design Kinds of data that belongs in DB
Data that requires updates from multiple users. Data that must accessed by multiple application programs. Data that require updates via transaction. Data that are long-lived and highly valuable. Data that must be secure.

System Design Note: Most application need a database, you should use relational DBMS. RDBMS features are sufficient for most application.

Handling Global Resources
System Design Handling Global Resources System designer must identify Global resources Mechanism for controlling the access. There are several kinds of global resources. Physical units : Processors, tape drives and communication satellites. Space: Disk space, workstation screen and buttons on a mouse

System Design Resource can be
Logical names: Object IDs, Filenames and class names. Access to shared data: Database Resource can be Physical object : It can control itself by specifying protocol Logical object: Conflicting access in a shared environment. To solve conflicting access, you should introduce “Guardian object”.

System Design Guardian object - own each global logical resource and have control access to it. All access must be pass through Guardian object only In ATM example, Bank code and Account numbers are global resources. Bank codes must be unique within consortium. Account number must be unique within the Bank.

Choose the Software control strategy
System Design Choose the Software control strategy Two kinds of control system in Software system: External control Internal control External control concerns the flow on externally visible events among objects. Three kinds of control external event: Procedure-driven sequential Event-driven sequential Concurrent

System Design Control style depend on the available resource (language, OS) and on the kind of interaction. Internal control refers to the flow of control within the process. It is only used in implementation therefore neither concurrent or sequential.

Procedure-driven Control
System Design Procedure-driven Control Control resided inside the program code. Basic working: Procedure request for input event and wait for it. When event arrives, control resumes within the procedure that made the call. Advantage: Easy to implement with conventional programming.

System Design Procedure-driven paradigm is suitable only if state model shows a regular alternation of input and output events. Note: C++ and Java are procedural languages. That is why they fail to support concurrency inherent in objects.

System Design Event-driven Control
Control resides within a dispatcher or monitor that the language, subsystem or OS provides. Basic Working Dispatcher call the procedure when events occurs. Procedure call to the dispatcher sent output or enable input but do not wait for it in-line.

System Design Once event is over, procedure return control to the dispatcher instead on retaining. Advantage: More flexible control than procedure-driven systems. Event-driven system are more modular and can handle error conditions better than procedure-driven system.

System Design Concurrent Control
Control resides concurrently in several independent objects, each a separate task. Basic Working: A task can wait for input but other tasks continue execution. OS resolves scheduling conflicts among tasks and usually supplies a queuing mechanism so that events are not lost.

System Design Internal control
Internal object interaction are similar to external object interaction because you can use the same implementation mechanisms. Importance difference is External interaction inherently involve events, because objects are independent. Internal interaction involve operation as procedure calls.

System Design Other paradigm are possible such as
Rule based systems Logic programming system Other forms Nonprocedural programs Developer used such languages in limited areas only such as Artificial Intelligence Knowledge based programming

Handling Boundary Conditions
System Design Handling Boundary Conditions Consider boundary conditions as well and address kinds of issues. Initialization System must initialize constant data, parameters, global variables, guardian objects, and possibly the class hierarchy. Initialize concurrent task is most difficult because object are independent.

System Design Termination Failure:
Termination is simpler than initialization because internal object can be abandoned. In concurrent system, one task must notify other task of its termination. Failure: Failure is unplanned termination of a system. Basically it arise from user errors or from external breakdown. It can also arise from bugs in the system

Common Architectural Styles
System Design Common Architectural Styles Several kinds of system listed : Batch transformation Continuous transformation Interactive interface Dynamic simulation Real-time system Transaction Manager

System Design Batch transformation
A batch transformation performs sequential computations. Working: Application receives the input Compute the answer No ongoing interaction with outside world Example: Computational problems such as compilers, payroll processing etc.

Determine connectivity
System Design For batch transformation problem, we can use class and interaction models. As it contains input, output and intervening stages. Most important thing is it should define a clean series of steps. Parse Text Determine connectivity Abstract to OO model Generate DB code ASCII code Graphics model Connectivity model class model Database model

System Design Compiler has five class model
One for input, one for output and three for intermediate representation. Steps for designing batch transformation: Break overall transformation into stages Each stage performing one part of transformation. Prepare class models for input, output and intermediate stages. Expand each stage until the operation are easy to implement. Restructure the final pipeline for optimizations.

Continuous transformation
System Design Continuous transformation It is a system in which the outputs actively depends on changing inputs. In batch transformation, output compute one time. In continuous transformation, updates output frequently. System can’t recomputed each set of output each time an input changes.

System Design Implement continuous transformation with a pipeline of functions. Pipeline propagates the effect of each input change. To improve the performance of pipeline, define intermediate and redundant objects.

Steps for designing a pipeline for continuous transformation
System Design Steps for designing a pipeline for continuous transformation Break overall transformation into stages Each stage performing one part of transformation. Define input, output and intermediate models between each pair of stages. Differentiate each operation – that is propagate the incremental effects of each change to an input through the pipeline as series of incremental updates. Add additional intermediate objects for optimization

Interactive Interface
System Design Interactive Interface System that is dominated by interaction between the system and external agents, such as humans or devices. System can not control the agents. Example of interactive system include forms-based query interface, Workstation windowing system and Control panel for a simulation.

System Design Major concerns of interactive interface are
Communication protocol between system and agents Syntax of possible interaction Presentation of output Flow of control within the system Performance Error handling

Steps for designing interactive
System Design Steps for designing interactive Separate interface classes from the application classes. Use predefined classes to interact with external agents Use the state model as the structure of the program.( i.e. concurrent, event-driven or procedure-driven control) Isolate physical events from logical event. Fully specify the application functions that invoked by the interface.

System Design Dynamic Simulation
A dynamic simulation models or tracks real-world objects. Ex. Economic models, video games. Objects and operation com directly from the application. Two ways for implementing control: Explicit controller to application object Objects can exchange messages among themselves.

Steps in designing dynamic simulation
System Design Steps in designing dynamic simulation Identify active real-world objects from the class model. Identify discrete events. Identify continuous dependencies. (i.e. one attributes may depend on other) Discrete events between objects can be exchanged as a part of timing loop.

System Design Real-time system
A real-time system is an interactive system with tight time constraints on actions. Two types of real-time system Hard real-time system Critical application that require a guaranteed response within the time constraints. Soft real-time system Highly reliable, but occasionally violate time constraints.

System Design Real-time design is complex and involves issue such as
Interrupt Handling Prioritization of tasks Coordinating multiple CPUs

System Design Transaction manager
Main function is to store and retrieve data. It deal with multiple users who read and write data at the same time. It also secure data from unauthorized access. It is built on top of a DBMS

System Design Steps in designing
Map the class model to database structure. Determine the units of concurrency. Determine the units of transaction. Design concurrency control for transaction.

Topic Covered Application Interaction Model Application Class Model Application State Model Adding Oerations

Application Interaction Model
Most domain model focus on building a model of intrinsic concepts. While Application model focus on the details of the application and consider interaction. You can construct application interaction model with following steps: Determine the system boundary Find actor Find use cases

Find initial and final events Prepare normal scenarios Add variation and exception scenarios Find external events. Prepare activity diagram for complex use cases. Organize actors and use cases Check against the domain class model.

Determine the system boundary Must know scope of an application to specify functionality. It means, you must decide what system includes and what it omits. If boundary is correct, you can treat system as box where internal details are hidden and changeable. At this state, determine purpose of the system.

Don’t consider humans as part of system, Ex. From problem statement (Chap 11), mentioned “ design system for human cashiers and Automatic Teller machine (ATM)..” So here there will two different system is going to be design. Human Cashier (Will be used at Bank) ATM ( At ATM location). Note: Will focus on ATM behavior only.

Finding Actors Identify the external object that interact directly with the system called Actors Actors includes Humans, external devices and other software systems. In finding actors, we are not searching for individual but for standard behavior. Each actor should be idealized.

It is possible for different kinds of external to play the same actor. Ex. ATM application, the actors are Customer, Bank and Consortium.

Finding Use Cases For each actor, list the functionality different ways in which the actor uses the system called Use Cases. Use cases partition the functionality of a system into a small number of discrete units. Each use cases should represent a kind of service that system provides – something that provides value to the actor.

Keep similar level of detail for use cases “Apply for loan” It should not be “Withdraw cash from saving account using ATM” . Restate with “Make Withdrawal” Now draw preliminary use case diagram. Show actors and use cases, connect actors to use cases. Usually, you can associate a use case with the actor that initiates it.

You should also write a one or more sentence for each use case.

Initiate session: ATM establish the identity of the user and make a list of accounts and actions. Query Account: System provides general data for an account, such as current balance, date of last transaction etc. Process Transaction: transaction like deposit, withdraw and transfer. Transmit Data: ATM uses the consortium’s facilities to communicate with the appropriate bank computers.

Finding Initial and Final Events Use case diagram does not show behavior clearly. To Understand behavior, you must understand the execution sequences of each use cases. Determine which actor initiate the use case. In may case, initial event is request for services that use case provides.

In many cases, initial event is occurrence that triggers a chain of activity. Similar, determine final event(s). For ex. For “apply of loan” would continue until Application submit Loan grant or reject Loan is delivered. Paid off and Closed. User must define the scope for termination.

For ATM Example: Initiate session: Initial Event is customer’s insertion of a cash card. Two Final Event: system keeps cash card and system returns the cash card. Query Account: Initial Event is customer’s request for account data. Final event is system retrieve data for customer.

Process Transaction: Initial event is customer’s initiation of transaction. Two final event : committing or aborting it. Transmit Data: Initial event: customer’s request for account data. Recovery from network, power or kind of failure. Final event: successful transmission of data.

Preparing Normal Scenarios For each use case, prepare one or more story base scenarios. Scenario illustrate the major interactions, external display and information exchange. Scenario is a sequence of events among a set of interacting objects. Think in terms of sample interactions. Sometime problem statement describes full interaction sequences, but most of time you will have invent.

So prepare scenarios for “normal” cases – interaction without any unusual inputs or error conditions. Information values exchanged are event parameters. Ex. Entered password has password value as a parameter. So for each event, identify the actor that caused the event and the parameters of the events.

Adding variations and Exception Scenarios Once normal scenarios prepared, consider “Special” cases, such as omitted input, maximum and minimum values and repeated values. Then consider “error” cases, including invalid values and failures to respond. Consider various other kind of interactions also such as help request and status inquiries.

ATM Example: ATM can’t read the card Card has expired ATM times out waiting for a response Amount is invalid Machine is out of cash or paper. Communication lines are down. Transaction rejected because of suspicious patterns of card usage.

Finding External Events To find all external events – includes all inputs, decisions, interrupts, and interaction to or from users or external devices. An Event can execute effects for a target object. Use scenario to find out normal events, unusual event and error conditions. In simple words, transmission of information to an object is Event.

For ex. Enter password is message from external User to application object ATM. Event instances whose values affect the flow of control should be different kinds of events. Account OK, Bad Account and Bad Password are different events. Based on event entered, prepared sequence diagram for it.

Prepare a sequence diagram for each scenario. It shows participation in an interaction and sequences of message among them. From sequence diagram, you can then summarize the events that each class sends and receives.

Sequence diagram for the process Transaction Scenario

Preparing Activity diagrams for
Application Interaction Model Preparing Activity diagrams for Complex use Cases Sequence diagram capture dialog and interplay between actors. Do not clearly show alternatives and decisions. Activity diagram let you consolidate all the behavior by documenting forks and merges in the control flow.

Organizing Actors and Use cases Next step to organize use cases with relationship (includes, extend and generalization) It will helpful for large and complex systems. For. Admin might be an operator with additional privileges.

Checking Against the Domain Class model Application and domain models should be mostly consistent. The actors, use cases and scenarios are all based on classes and concept from domain model. Cross check the application and domain models to ensure that there are no inconsistencies.

Application Class Model
Construct an application class model with following steps: Specify User Interfaces Define Boundary classes. Determine controllers. Check against the interaction model.

Specifying User Interface Most interaction divided into two parts Application Logic User Interface A user interface provides the user with way to access its objects, command (function/Features) and application options. Same program logic can accept input from command lines, files, mouse buttons, touch pane, physical push buttons, or remote links.

It is acceptable to sketch out a sample interface to help you visualize the operation of an application. Might need mock up the interface so that user can try it. Dummy procedure can simulate application logic. It will help you to evaluate the “look and feel” of the user interface.

Defining Boundary Classes It is always helpful to define boundary classes to isolate the inside of a system from the external world. A boundary class is a class that provides a staging area for communication between a system and an external source. It understand the format of one or more external sources and converts information for transmission to and from the internal system.

For ex. To summarize the communication between ATM and consortium, we need to define boundary classes (CashCardBoundary, AccountBoundary).

Determining Controllers A controller is active object that manages control within the application. It receives signal from outside world. Reacts to them Invokes operation on the objects in the system. Send signals to outside world. A controller is piece of reified behavior captured in form of object.

For Ex. ATM has two major control loops. Outer loop verifies customer and account. Inner loop services transactions.

Checking against the Interaction Model Once you build class model, go over the use cases and think about how they would work. For Examples user sends a command Parameters of command must come from UI object. Requesting a command itself must come from some controller. Finally, simulate use case with the classes.

Application State Model
Application state model focuses on application classes and enhances the domain state model. Application class are more important than domain classes. Identify application classes with multiple states and use the interaction model to find events for these classes. Then organized event sequences for each class with a state diagram.

Next check various state diagram to make sure that common events match. Finally check state diagram against the class and interaction model. So to construct application state model with following steps: Determine application classes with states. Find events Build state diagram

Check against other state diagram Check against the class model Check against the interaction model.

Determining Application Classes with States Identify classes that are prominent to users and important to the operation. Consider each application class and determine which one have multiple states. User interface classes and controller classes are good candidate. In contrast, boundary classes are less likely to be involved in state model.

Finding Events From prepared scenario, extract events. Don’t overlook common interaction and highlights the major events. Difference between domain and application processes for state model. With Domain model, first we find states and then we find events. Because its focus on data. With Application model, first we find events and then states. Ex. Insert card, enter password etc.

Building State Diagram Next step, build state diagram for each application class. Select any one class and consider a sequence diagram for it. Now find event from that. Interval between any two event is a state. Give each state a name. Now merge other sequence diagrams into the state diagram.

Now finds loops within the diagram. If a sequence of events can be repeated indefinitely then they form a loop. Once loops found, merge other sequence diagram and look for alternative path. Another alternative is to partition a state diagram into two concurrent sub-diagram, using one subdiagram for the main line and other for distinguishing information.

Once normal events considered, add variation and exception cases. When diagram covers all scenarios, you are finished with state diagram. Now refer each state diagram figure. for ATM Example.

Checking against other state diagram Check state diagram for each class. Every event should have sender and receiver. Follow the effects of input event from object to object to make sure that they match scenarios. For Ex. SessionController initiates the TransactionController, and the termination of the TransactionController causes the SessionController to resume.

Application State Model Checking against the Class Model
Same way, make sure that the state diagram are consistent with the domain and application class models. ATM Example: Multiple ATMs concurrently access an account. And account access need to be controlled to ensure that only one update a time is applied.

Checking against the Interaction Model When state model is ready, go back and check it against the scenarios of interaction model. Simulate each behavior sequence by hand and verify that state diagram give correct behavior. If error discovered, change state diagram or scenarios. Don’t assume that the scenarios are always correct.

Adding Operations Till now we did not focus on operations because the list of useful open-ended and it is difficult to know when to stop adding them. Following are source through which we can add operations: Operations from Class model , Operations from Use Cases, Shopping-List Operations and Simplifying operations.

Operation from Class model: To add operation, note down attributes values and links implies by Class model because it does not shows explicitly. Operation from Use Cases: To add operation, we are using use cases. Most complex functionality of system comes from use cases.

Shopping list operations: Shopping list mean Operations are not dependent on particular application but are meaningful in their own right. ATM Exam: Bank.createSavingAccount() Bank.CreateCheckingAccount() AddAccount() Remove Account() Close()

Simplifying Operations: Study class model for similar operation and variations in form on a single operations. Use inheritance where possible to reduce the number of distinct operations. If needed then only introduce New superclass. Locate each operation at correct level within the class hierarchy.

Chapter 12 Domain Analysis
Prepared By: Ashwin Raiyani

Topic Covered Introduction Overview of Analysis Domain Class Model Domain State Model Domain Interaction Model

Introduction During analysis, we build models and begin to understand the requirements deeply. To build a domain model, you must interview business expert, examine requirements statements, and study related requirements. Successful analysis model states what must be done, without restricting how it is done and avoid implementation decisions.

Overview of Analysis

Overview of Analysis As fig. shows, analysis begin with problem statement during system conception. Problem statement can be incomplete or informal but analysis makes it more precise and expose ambiguities. You must understand real-world system described by the problem statement, and abstract its essential features into a model. Sequence can be problem statement  build model (Domain)  build model(application).

Overview of Analysis Analysis model addresses the three aspects of objects. Static structure of objects (Class Model) Interaction among objects ( Interaction Model) Life-cycle histories of objects (State Model)

Overview of Analysis First step in analyzing the requirements is to construct a domain model. Static structure of the real world system is captured. The domain model describes real-world classes and their relationships to each other.

Domain Class Model Information for the domain model comes from the
Problem statement, Artifacts from related systems, Expert knowledge of the application domain and General knowledge of the real world. The steps to be performed to construct a domain class model: Find Classes. Prepare a data dictionary.

Domain Class Model Find associations.
Find attributes of objects and links. Organize and simplify classes using inheritance. Verify that access paths exist for likely queries. Iterate and refine the model. Reconsider the level of abstraction. Group classes into packages

Domain Class Model 1. Finding Classes
First Step, find relevant classes for objects from application domain. It includes houses, person, machines etc. Classes often correspond to nouns. Eg- ” a reservation system sell tickets to performances at various theater”- Tentative classes would be Reservation, System, Tickets, Performance and Theaters.

Domain Class Model Idea is to capture concepts. not all nouns are concepts, and concepts are also expressed in other parts of speech. For the Case study of the ATM: The following are the classes extracted from problem statement nouns.

Domain Class Model

Domain Class Model Additional classes that do not appear directly in the statement but can be identified from our knowledge of the problem domain

2.Keeping the Right classes
Domain Class Model 2.Keeping the Right classes Discard unnecessary and incorrect classes according to the following criteria. Redundant classes: If two classes express the same concept, you should keep the most descriptive name. ATM example. Customer and user are redundant; we retain customer because it is more descriptive.

Domain Class Model Irrelevant classes: If class has little or nothing do with application, eliminate it. ATM Ex. cost is outside the scope of the ATM software. Vague classes: class should be specific. ATM Example, System, Security provision, Banking network etc are not specific thing. Attributes: Names that primarily describe individual objects should be restated as attributes.

Domain Class Model ATM Example, Account Data, Cash, Transaction data are purely indicating attributes not a class. Operations: If a name describes an operation that is applied to objects and not manipulated in its own right, then it is not a class. Eg-if we are simply building telephones, then call is part of the state model and not a class

Domain Class Model But Billings system for telephone calls a Call would be important class with attributes date, time, origin and destination. Roles: The name of a class should reflect its intrinsic nature and not a role that it plays in an association. Ex.Owner of a car in a car manufacturing database, not correct as a class. It can be a person( owner, driver, lessee)

Domain Class Model Implementation Constructs: Eliminate constructs from the analysis model that are irrelevant to the real world. We may need them during design and not now. Ex. Transaction Log class. Derived classes: As a general rule, omit classes that can be derived from other classes. Mark all derived classes with a preceding slash(‘/’)in the class name.

3. Preparing a Data Dictionary
Domain Class Model 3. Preparing a Data Dictionary Prepare a data dictionary for all modeling elements. Describe the scope of the class within the current problem, including all assumptions or restrictions on its use. It also describes associations, attributes, operations and enumeration values.

Domain Class Model Data Dictionary for the ATM classes Account, ATM,
Bank, BankComputer, CashCard, Cashier, CashierStation CentralComputer, Consortium, Customer, Transaction

Domain Class Model 4. Finding Associations
Find A structural relationship between two or more classes is an association. A reference from one class to another is an association. Associations often correspond to verbs or verb phrases. Ex. Physical Location ( part of, NextTo) Directed Actions (Drives) Communication (Talks To)

Domain Class Model Idea here is to capture relationships
Ownership ( Has, Part of) Satisfaction of condition ( WorksFor, Manages). Idea here is to capture relationships

Domain Class Model

5. Keeping the Right Associations
Domain Class Model 5. Keeping the Right Associations Discard unnecessary and incorrect associations, using the following criteria: Associations between eliminated classes: If you have eliminated one of classes then either you eliminate association or restate it. Ex. Banking Network includes cashier stations and ATMs. ATM dispenses cash

Domain Class Model ATM prints receipts Bank provide software Cost apportioned to banks System provides record keeping and System provides security. Irrelevant or implementation associations: Eliminate any association that deals with implementation or outer problem statement. Ex. System handles concurrent access (Implementation)

Domain Class Model Actions: An association should describe a structural property of the application domain not a transient event. Ex. ATM accepts Cash card (Interaction cycle) It is not a permanent relationship between ATM and Cash. Eliminate ATM interact with user. Central computer clears transactions with bank .

Domain Class Model Ternary associations: You can decompose most association among three or more classes into binary associations. Always decomposed without losing information. Ex. Bank computer processes transaction against account can be convert into Bank computer processes transaction and transaction concern accounts.

Domain Class Model Derived associations : Omit association and attributes, they may be redundant. Ex. GrandParentOf can be defined in terms of pair of ParentOf. Ex. youngerThan expresses condition on the birthdate of two person, not additional information. Derived association don’t add information, they useful for understanding.

Semantics of Association
Domain Class Model Semantics of Association Misnamed Association: Name are important to understanding and should be chosen with care. Ex. Bank computer maintain accounts. Rephrase as Bank hold account. Association End name: Add association end name where appropriate. Ex. Person manages person. It would be appropriate to give end names boss and worker

Domain Class Model Qualified Associations: Most names are not globally unique. So context combines with the name to uniquely identify the object. Ex. Company name unique within one state but may be duplicated in other state. So combining State + Company name will uniquely identify company. Ex. bankCode differentiate bank in a consortium.

Domain Class Model Multiplicity: Don’t put much effort as it is often changes during analysis. Missing Association: Add any missing association that are found during analysis. Transaction entered on cashier station, Customers have accounts and transaction authorized by cash cards. To perform above, we need to introduce relationship from Cashier to cashier station. So association Cashier authorized on cashier station needed

Domain Class Model Aggregation: it is specially for mechanical parts or bills of material. Don’t spend much time trying to defer between association and aggregation. Ex. Bank is part of Consortium and indicate the relationship with aggregation. Now combining all things together, class diagram prepare.

Domain Class Model

Domain Class Model 6. Finding Attributes:
Attributes are data properties of objects like colour, weight etc. Attributes usually correspond to nouns followed by possessive phrases, such as “the color of the car” Attributes are less likely to be fully described in problem statement. Only consider attributes directly relevant to application. Get important attributes then add details to it.

Domain Class Model Avoid derived attributes.
EX. Age is derived from birthdate and currentTime Looks for attributes on associations. Ex. Workfor association attribute can be salary, title etc.

7. Keeping the Right Attributes
Domain Class Model 7. Keeping the Right Attributes Eliminate unnecessary and incorrect attributed with the following criteria: Objects: if element is important rather than just its value, then it is an object. Ex. Boss refers to a class and Salary is an attributes. Name: Name often refer as Qualifier rather than attributes. Name is an attribute when its use does not depend on context,

Domain Class Model Ex. Names of person are not unique therefore its attributes. Identifiers: it mean referencing objects used for some work. Ex. ATM transaction always generate Transaction ID for each operation. So you can count Transaction ID as attributes. Attributes on Association: If value require the presence of link then attributes of the association should derived.

Domain Class Model Ex. Workfor association attribute can be salary, title etc Internal Values: if an attributes describes internal state of an object then eliminate it. Fine Detail: Omit minor attributes that are unlikely to affect most operation. Discordant attributes :- an attribute that seems completely different from and unrelated to all other attributes then remove it.

Domain Class Model Boolean Attributes: Convert Boolean attributes into enumeration.

Domain Class Model

8. Refining with Inheritance
Domain Class Model 8. Refining with Inheritance Next step, to share common structure. Apply in two directions: Bottom Up By Generalizing common aspects of existing classes into a superclass. Top Down By Specializing existing classes into multiple classes

Domain Class Model Bottom Up Generalization:
Searching for classes(from bottom up) with similar attributes, associations and operation. For each generalization, define a superclass to share common features. May have to slightly redefine some attributes or classes to fit in. But don’t push too hard it create wrong generalization. Ex. RemoteTransactiona and CashierTransaction are similar and can be generalized by Transaction.

Domain Class Model It main derived from application domain itself.
Top-Down Specialization: It main derived from application domain itself. Look for noun phrases composed of adjectives: Fixed menu, sliding menu and text menu. Avoid excessive refinement.

Domain Class Model Generalization Vs. enumeration :-
Generalization is all about common structure Enumeration is all about list of values. Ex. CurrentAccount and SavingAccount share common structure but it does not affect behavior within the ATM application. So Type can introduces as attributes of account and enumerate it. Multiple Inheritance: if require then apply because it increasing both conceptual and implementation complexity

Domain Class Model Similar association:
when the same association name appears more than once with the same meaning, try to generalize the associated class. Ex. EntryStation generalizes CashierStation and ATM. Adjusting inheritance Level Assign attributes and association to specific classes in the class hierarchy. You may need some adjustment to get everything right.

Domain Class Model 9.Testing Access Paths
Verify that access paths exist for likely queries. Trace access paths through the class model to see if they yield sensible results. Make sure you have not overlooked any associations.

10. Iterating a Class model
Domain Class Model 10. Iterating a Class model A class model is rarely correct after a single pass. If you find any deficiency, go back to an earlier stage if necessary to correct it. So iterate and refine the model. There are some sing of missing classes. Asymmetries in association and generalization. Disparate attributes and operation on a class. Difficulty in generalizing cleanly.

Domain Class Model Look out for missing associations
Duplicate association with same name and purpose. A role that substantially shapes the semantics of a class. Ex. It mean converting association into a class. Look out for missing associations Missing access paths for operations Lack of attributes, operations and association on a class. Redundant information: Adjust the placement of attributes and associations

11.Shifting the level of abstraction
Domain Class Model 11.Shifting the level of abstraction Abstraction makes a model more complex but can increase flexibility and reduce the number of classes. In case of abstraction, we need to think in terms of pattern. A pattern distills the knowledge of experts and provide a proven solutions to a general problem. Ex. Management hierarchy.

Domain Class Model

12.Group classes into packages.
Domain Class Model 12.Group classes into packages. The last step of class modeling is to group classes into packages. A package is a group of elements(classes, association, generalizations and lesser packages) with common theme. When you place classes and association in a package, you are making semantic statement.

Domain Class Model Therefore, package might be:
Tellers – Cashier, Entry Station, Cashier Station, ATM Accounts – Account, cash card, card authorization, customer, transactions, update, cashier transaction, remote transaction. Bank- consortium, bank Each package can add details to it.

Domain State Model The Following steps are performed in constructing a domain state model Identifying classes with states Finding states Finding Events Building state diagrams Evaluating state diagrams

1. Identifying Classes with states
Domain State Model 1. Identifying Classes with states Study list of domain classes. Look for classes that can be characterized by a progressive history or represent cyclic behavior. Identify significant states in the life cycle of an Object. Not every state occurs in every cycle. ATM Example, Account is appropriate behavior for ATM. Life cycle of Account is progressive and cycling to and from problem states.

Domain State Model Finding States List the state for each class.
Characterized the object by their Attributes Values Associations that may participants Attributes and association that are meaningful in certain states only. Avoid names that indicate how the state came.

Domain State Model By looking at events and considering transitions among states, missing states will become clear. Ex. Some states for Account Normal (Normal access) Closed ( Closed by customer) Overdrawn ( withdrawal exceeds the balance) Suspended ( blocked for some reason)

Domain State Model Finding Events
Now find events that cause transitions among states. Think about stimuli (input) that cause a state to change. Find other events that takes object into a specific states Ex. Pickup receiver on telephone, it enters into Dialing State. But many telephone has pushbuttons that invoke specific functions.

Domain State Model So there are some additional event that occur within a state and don’t cause a transitions. For Domain State model, make focus on events that cause transition among states. Ex. Event includes: close account, withdraw excess funds, repeated incorrect PIN, suspected fraud and Administrative action.

Building State diagrams
Domain State Model Building State diagrams Determine for which state, each event applies. Add transitions to show the change in state caused by the occurrence of an event when an object is in particular state. Once you have specified the transitions, check does it represent an error or not? If yes then add transitions to error state.

Domain State Model

Evaluating State Diagram
Domain State Model Evaluating State Diagram Examine each state model. Are all states connected? Path from initial state to the final state? Are the expected variations represent it? Are there any dead states that terminate the cycle? Find missing path and states from it. When complete, it should indicate life cycle of the class.

Chapter Process Overview
Prepared By: Ashwin Raiyani

Development Life Cycle
Topic Covered Development Stages System Conception Analysis System Design & Class Design Implementation & Testing Training, Deployment and Maintenance Development Life Cycle Waterfall Development Iterative Development

Development Stages System Conception
System Conception deals with origin of the system or application. S/W development begins with - business analyst or users conceiving an application & formulating tentative requirement.

Focuses on creation of models.
Development Stages Analysis Focuses on creation of models. It specify What must be done, not how it should be done. Task of Analyst Must work with the requester (client) to understand the problem, because problem statement are rarely complete or correct. There are two stages of analysis: Domain Analysis & Application Analysis

Development Stages Domain Analysis Focus on real-world things whose semantics the application captures Ex. Airplane Flight is real-world object, Flight reservation system must represent it. Domain model is mainly to decide which information to capture and how to represent it.

Application analysis Development Stages
Focuses the computer aspects of the application that are visible to users. Ex. Flight reservation screen is part of flight reservation system. It does describes the application appears from the outside. It does not describe the implementation of the application

Development Stages System Design During system design, developer makes strategic decisions relevant to system architecture. Architecture is high level plan or strategy for solving application problem. Architecture is depend on requirement and based on past experience. Must understand how a new system interacts with other systems.

Add details to analysis model
Development Stages Class Designer Task of class designer Add details to analysis model They determine data structures & algorithm for each of the operation of window class. They elaborate both domain & application objects using same OO concept & notation.

Development Stages Implementation It is the stage of writing code. Translating design elements into programming language and database code. Tools can used to generate code.

Testing Development Stages
Now system is ready but it must be carefully tested before delivered to customer. Tester once again revisit the original requirement and verify the proper functionality. Can discover errors or bugs. Tester perform testing on all hardware and software platform

Training Development Stages
Organization must train users so that they can get full benefit from an application. Team should prepared separate user documentation(i.e. Help file or user manual)

Deployment Development Stages Deployment mean, product release
Once training is over, system is ready to install. System must work on various platform and on various configuration. It should support multiple languages.

Maintenance Development Stages
Bugs that remain in the original system will gradually appear during use and must be fixed. Preserve the long-term viability of the application.

Software Development Life Cycle
Waterfall Development In this approach, developer has to follow strictly linear sequence with no backtracking. First, capture requirement, construct analysis model then prepare system design, class design and followed by implementation, testing, deployment. Suitable for well understood applications with predictable outputs from analysis and design, such systems seldom occur.

A waterfall approach also does not deliver a useful system until completion. Drawback : This makes it difficult to assess progress and correct a project that has gone wrong.

Iterative Development
Software Development Life Cycle Iterative Development More flexible. There are multiple iterations as the system evolves to final deliverable. Each iteration includes a full complement of stages: analysis, design, Implementation and testing. This is the best choice for most applications because it gracefully responds to changes and minimizes risk of failure.

Management and business users get early feedback about progress.

<<enumeration>>
An enumeration is a data type that has finite set of values: For ex. Enumeration often occur and are important to users. i.e. it display only the possible values or users are restricted to access only allowed values or data. Do not use generalization to capture the values of an enumerated attributes <<enumeration>> week Sunday Monday Tuesday ……. Saturday Calendar week: Week

<<enumeration>>
An enumeration  list of values Generalization  Structuring the description of objects. Ex. We are not introducing generalization for “Card” Class because most games do not differentiate the behavior of “Suit”. <<enumeration>> Suit Spades Clubs Hearts diamonds Card suit:suit rank: rank

UML notation Top section: Write keyword enumeration in between “<<>>” above enumeration name. Second section : List the enumeration values.

Multiplicity Multiplicity is a constraint on the cardinality of a set.
Multiplicity for an attribute specifies the number of possible values for each representation of attributes. Common specifications are Mandatory single value [1] An optional single value [0..1] Many[*] It specify whether an attribute is mandatory or optional It indicates if an attribute is single valued or can be a collection.

If not specified, it is assumed to be a mandatory single value[1]
Person name: string[1] address:string[1..*] birthdate:date[1]

Scope It indicates if a feature applies to an object or a class.
You should avoid attributes with class scope because they can lead inferior model. It is better to model group explicitly and assign attributes to them. PhoneMessage source Phon Box maximumDuration maxDaysRetained dateRecorded timeRecorded priority Person name 0..1 * 1 * maxMessageCount phoneNumber password owner {ordered} 1 *

Upper model is inferior because the maximum duration, maximum day retained and maximum message count have a single value for the entire phone mail system. It is acceptable to use an attribute with class scope to hold the extent of a class ( Set of objects for a class) source PhoneMessage dateRecorded timeRecorded priority MailCategory maxMessageCount maximumDuration maxDaysRetained Phon Box phoneNumber password 0..1 * 1 * owner {ordered} 1 * * Person name owner 1

It is also acceptable to define operations of class scope.
It is convenient to define class scoped operations to provide summary data.

Visibility Visibility refers to the ability of a method to reference a feature from another class and has the possible values of Public, Protected Private Package Public features: Any method freely access Protected features: only methods of the containing class and its descendant via inheritance can access Private features: only methods of the containing class can access. Package features: Methods of classes defined in the same package as the target class can access.

UML denotes visibility with a prefix.
Character “+” precedes public Character “-” precedes private Character “#” precedes protected Character “~” precedes package Several issues when choosing visibility Comprehension You must understand all public features to understand the capabilities of a class. Others are merely an implementation convenience.

Extensibility Context
Many classes can depend on public methods, so it can be highly disruptive to change their signature. Fewer classes depend on private, protected, and package methods, there is more latitude to change them. Context Private, protected and package methods may rely on preconditions or state information created by other methods in the class. A private method may calculate incorrect results or cause the object to fail.

Association ends A binary association has two ends, a ternary association has three ends and so forth. Few properties: Association end name: An association end may have meaningful name. Names clarifies multiple references to a class and facilitate navigation Multiplicity: Can specify multiplicity for each association end. Ordering: Objects for a “many” association end are usually just a set. Bags & Sequences: Objects for a “many” association end also be a bag or sequence. Qualification: One or more qualifier attributes can disambiguate the objects for “many” association end

Additional properties
Aggregation: It may be an aggregate or constituent (element) part. Only binary association can be a aggregation: One association end must be an aggregate Other must be a constituent Changeability: It specifies the update status of an association end. Navigability UML shows navigability with an arrowhead on the association end attached to the target class. Arrowheads may be attached to zero, one or both ends of an association. Visibility Similar to attributes and operations, association ends may be public, protected, private or package.

N-ary Associations Project Language * * * programmer Person N-ary associations mean associations among three or more classes. Ex. Programmers use computer languages on projects. UML notation: It is diamond with lines connecting to related classes. If the association has a name, written in italics next to the diamond. Note: Avoid n-ary associations – most of them can be decomposed into binary associations, with possible qualifiers and attributes.

N-ary association can have a name for each end just like binary associations.
You can not traverse n-ary associations from one end to another as with binary associations. Professor * * * DeliveredCourse roomNumber Textbook Semester * * ListedCourse

Ex. A professor teaches a listed course during a semester
Ex. A professor teaches a listed course during a semester. The resulting delivered course has a room number and any number of textbooks. Programming language can not express n-ary associations. So you need to promote (support) n-ary associations to classes. When you promote n-ary associations to a class, might change the meaning of a model. An n-ary associations enforces that there is most one link for each combinations. Ex. For each combinations of professor, semester and ListedCourse there is one DeliveredCourse So if you were implementing below fig. special application code would have to enforce the uniqueness of Professor + Semester + ListedCourse.

Professor 1 * DeliveredCourse roomNumber 1 * * * Semester Textbook * 1 ListedCourse

UML has two forms of part-whole relationships:
Aggregation - A general form Composition – More restrictive.

Aggregation Definition:
An aggregation as relating an assembly class to one constituent part class. An assembly with many kinds of constituent parts corresponds to many aggregations. For ex. College consists of students, professor, courses etc. Here college is assembly and the others are constituents. Therefore, college to student is one aggregation, college to professor is another aggregation. We define each individual pairing as an aggregation. So that we can specify the multiplicity of each constituent part within the assembly.

Aggregation is an special form of binary associations.
Most important property of aggregation: Transitivity If A is part of B and B is part of C, then A is part of C. Antisymmetric If A is part of B, then B is not part of A. Aggregate operations imply transitive closure and operate on both direct and indirect parts.

Aggregation v/s Association
Aggregation is not an independent concept, it is a special form of association. If two objects are tightly bound by a part-whole relationship, it is an aggregation. If two objects are usually considered as independent and have linked, it is association. To find whether there is a aggregation or not, some tests includes: Use the phrase part of Some operations on the Whole apply to its parts Some attributes values circulate from whole to all or some parts.

Like association, except a small diamond indicates the assembly end.
UML notation: Like association, except a small diamond indicates the assembly end. Decision to use aggregation is Matter of Judgment Arbitrary College Courses Professor Student

Aggregation v/s Composition
Composition is a form of aggregation with two additional constraints. A constituent part can belong to at most one assembly. Once its assigned, it has coincident lifetime with the assembly. Composition implies ownership of the parts by the whole. Composition is convenient for programming Ex. Deletion of assembly object triggers deletion of all constituent objects.

Composition is a small solid diamond next to assembly class.
UML notation: Composition is a small solid diamond next to assembly class. EX. A company consists of divisions, which in turn consists of departments. A Company is indirectly a composition of departments. 1 * 1 * Company Division Department 1 WorksFor * Person

Propagation of operations
Propagation is the automatic application of an operation to a network of objects. When the operation is applied to some starting object. Ex. A persons owns multiple documents. Copy operations propagate from document to paragraph to characters. Copying a paragraph copies all the characters in it. Operation does not propagate in the reverse direction: A paragraph can be copied without copying the whole document.

UML notation: Small arrow indicating the direction and operation name next to the affected association. Note: this notation is not part of the UML and is a special notation.

Abstract Classes An abstract class is a class that has no direct instances but whose descendant classes have direct instances. A Concrete class is a class that is instantiable; it can have direct instances. Only concrete classes may be leaf classes in an inheritance tree.

Baker and candelstickMaker are concrete classes because they have direct instances.
Worker { note: listing of workers is incomplete} Baker CandlestickMaker Concrete Class: it can have direct instances.

FullTimeEmployee and PartTimeEmployee are concrete classes because they can be directly instantiated. UML Notation: Abstract class name listed in italic font or may place keyword {abstract} below or after the name. Abstract Class and Abstract operation: it has not direct instances.

An abstract class can define signature for an operation without supplying corresponding method.
An abstract operation defines the signature of an operation for which each concrete subclass must provide its own implementation. A concrete class may not contain abstract operations. UML Notation: An abstract operation is designated by italics or the keyword {abstract}. Note: Abstract nature of a class is always provisional, depending on the point of view or application.

Multiple Inheritance It permits a class to have more than one superclass and to inherit features from all parents. More complicated form of generalization than single inheritance. Advantage is greater power in specifying classes and an increased opportunity for reuse. Disadvantage is a loss of conceptual and implementation simplicity. Multiple inheritance mean either the conceptual relationship between classes or the languages mechanism that implements that relationship. Distinguish between Generalization : conceptual relationship Inheritance : Language mechanism.

Kinds of multiple Inheritance
Common form of multiple inheritance is from sets of disjoint classes. Each subclasses inherits from one class in each set. Employee employementStatus managerialStatus FullTimeEmp PartTimeEmp Manager IndividualContributor FullTimeIndividualContributor

Therefore, here class Employee has two aspect
Here, FullTimeEmp and PartTimeEmp are disjoint; and Manager and IndividualContributor are also disjoint classes. We can also define additional combinations classes here like FullTimeManager, PartTimeContributor, PartTimeManager and FullTimeIndividualContributor. Each generalization should cover a single aspect (A generalization set name). Therefore, here class Employee has two aspect Employment Status Managerial Status

Conflicts among parallel definitions create ambiguities.
So, FullTimeIndividualContributor inherits Employee features along two paths via employmentStatus and managerialStatus. Conflicts among parallel definitions create ambiguities. You should avoid such conflicts in models or explicitly resolve them. For Ex. FullTimeEmp and IndividualContributor both have attribute called name. FullTimeEmp.name could refer as person’s full name and IndividualContributor.name might refer as person’s title. Solution is restating the attributes of classes.

Multiple inheritance can also occur with overlapping classes.
landVehicle watervehicle Amphibiousvehicle boat Car Vehicle {overlapping, incomplete}

AmphibiousVehicle is both Land and water vehicle
AmphibiousVehicle is both Land and water vehicle. So there is a overlapping. UML uses constraint to indicate an overlapping generalizations set; Notation is a dotted line cutting across affected generalization with keyword in braces.

Multiple Classification
Definition: One instance happens to participant in two overlapping classes.

UML permits multiple classification but most OO languages handle it poorly.
Fig. show that Person as an object composed of multiple UniversityMember object. This workaround replaces inheritance with delegation. This is not totally satisfactory, there is a loss of identity separates roles.

Workaround Dealing with multiple inheritance is implementation issue but early restructuring of a model is often the easiest way to work around its absence. Two approaches make use of delegation, which is an implementation mechanism by which object forwards an operation to another object for execution. Restructuring technique Delegation using composition of parts Inherit the most important class and delegate the rest. Nested generalization

Delegation using composition of parts
You can recast superclass with multiple independent generalization as a composition in which each constituent part replaces a generalization. Single object having a unique ID by a group of related objects that compose an extended object. Inheritance of operations across the composition is not automatic. Composite must capture operations and delegate to appropriate part.

You need not create various combination as explicit classes.
Workaround for multiple inheritance - delegation You need not create various combination as explicit classes. All combinations of subclasses from the different generalizations are possible.

Inherit the most important class and delegate the rest.
Maintain identity and inheritance across the most important generalization. Reduce the remaining generalization to composition and delegate their operations. Workaround for multiple inheritance – inheritance and delegation

Nested Generalization:
Factor on one generalization first, then the other. This approaches multiplies out all possible combinations. This preserves inheritance but duplicates declarations and code and violates the spirit of OO programming.

There are several issue when selecting the best workaround.
Superclass of equal importance Dominant superclass Few subclass Sequencing generalization sets Large quantities of code Identity

Superclass of equal importance:
If a subclass has several superclasses it may be best to use delegation and preserves symmetry in the model. Dominant superclass: If one superclass clearly dominate and others are less important, preserve inheritance through this path. Few subclasses: If the number of combinations is small, consider nested generalization otherwise avoid. Sequencing generalization sets. In a case of generalization, factor on most important criterion first, then second most and so forth.

Large quantities of code:
Try to avoid nested generalization, in a case of duplicate large quantities of code. Identity: Consider the importance of maintaining strict identity. Only nested generalization preserves this.

Metadata Definition Metadata is data that describes other data.
For ex. A class definition is metadata. Models are inherently metadata, they describe the things being modeled.

A car model describes many physical cars and holds common data.
A car model is metadata relative to a physical car, which is data.

Constraints Definition
A constraint is Boolean condition involving model elements, such as objects, classes, attributes, links, associations and generalizations sets. A constraint restricts the values that elements can assume.

Constraints on objects
Ex-1. No employee’s salary can exceed the salary of the employee’s boss. ( a constraints between two things) Ex-2. No window can have an aspect ratio of less than 0.8 or greater than 1.5 ( a constraint between attributes). Ex-3. Priority of a job may not increase (a constraint on the same object) Employee.salary < boss.salary 0.8< length/width<1.5 Priority never increase

Constraints on Generalization sets
Class models capture many constraints through their structure. For ex. The semantics of generalization imply certain structural constraints. With single inheritance the subclasses are mutually exclusive UML had few keywords for generalization sets. Disjoint The subclasses are mutually exclusive. Each object belongs exactly one of subclasses. Overlapping The subclasses can share some objects. An object may belongs to more than one subclass. Complete: The generalization lists all the possible subclasses. Incomplete: The generalization may be missing some subclasses.

Constraints on Links Multiplicity is a constraint on cardinality set.
Multiplicity for an associations restricted the number of objects related to a given object. Qualification also constraints an association. A qualifier attribute does not merely describe the association but also significant in resolving the “many” objects. An association class implies a constraint. An association class has constraint that an ordinary class does not; it derives identity from instances of the related classes.

There is no particular order on the objects of a “many” end for ordinary association.
But, constraint {ordered} indicates that elements of a “many” associations end have explicit order.

Use of constraints Declaration lets you express a constraint’s intent, without supposing an implementation. That mean, convert constraints to procedural form before implementation. Practically, you can not enforces every constraint with model’s structure, but you should try to enforce the important ones.

You should try to position constraint near affected elements.
UML notation Two alternative notations of constraints: Delimit a constraint with braces or place it in a “dog-eared” comment box. You should try to position constraint near affected elements. A dashed arrow can connect a constrained element to the element on which it depends.

The subject is not primarily about OO-language or coding.
OOAD Concept The subject is not primarily about OO-language or coding. This subject emphasize on initial stages of process development i.e. Requirement Gathering Planning etc. Object-oriented analysis and design (OOAD) is a software engineering approach that models a system as a group of interacting objects.

OOAD is divided into two parts as follows:
OOAD Concept OOAD is divided into two parts as follows: OOA (Object Oriented Analysis): Definition :Object Oriented Analysis (OOA) is concerned with developing requirements and specifications OOD (Object Oriented Design) Definition: Object Oriented Design (OOD) is concerned with developing object-oriented models of a software/system to implement the requirements identified during OOA

What is an object? What is an object?
1) any concept that represent a single thing 2) a representation of a specific entity in the real world 3) may be tangible (physical entity) or intangible Examples: In the case of LMS, software or system itself is intangible object and student and Library clerk act as tangible objects.

What is Object-Orientation?
Organize software as a collection of discrete objects that incorporate both data structure and behavior. Data structure  data pattern or data style or formation of data Behavior  things that the object can do that are relevant to model OO approach has 4 characteristics: Identity Classification Inheritance Polymorphism

Characteristics of Objects
Identity Classification Polymorphism Inheritance

Characteristics of Objects - Identity
Definition : Discrete (i.e. unique) & distinguishable entities called objects. Ex. A person is an object. His name is Identify Ex. Ram Note: Objects are uniquely identifiable by name. Each objects has its own inherent identity. In other words, two objects are distinct even if an their attribute values (i.e. name and size) are identical. Ex. Twins are two distinct person by identity

Characteristics of Objects - Classification
Definition: Objects with the same data structure (attributes) & behavior ( operation) are grouped into a class. In OO systems, class is a set of objects that share common structure and common behavior Classes are important mechanisms for classifying objects. Each class describes a possibly infinite set of individual objects.

Characteristics of Objects - Classification
Each objects is said to be instance of its class. Objects has its own value for each attributes but shares the attributes names & operations. Example : Class Name : Circle Attributes : radius, center Operation : setCenter(), setRadius()

Characteristics of Objects - Inheritance
Definition: Sharing of attributes & operations (features) among classes based on hierarchical relationship. A superclass has general information that subclass refine and elaborate. Each subclass incorporates all the features of its superclass and adds its own features. In other words, defining new classes from the existing one.

Characteristics of Objects - Inheritance
Note: subclasses need not repeat the features of the superclass. Advantage: common features of several classes into a superclass can reduce repetition within design and programs.

Characteristics of Objects – Example of Inheritance
Inheritance is implied by is-a or kind-of relationship. class Person { String name; String age; void birthday () { age = age + 1; } class Employee extends Person { double salary; void pay () { ...} } Person Every Employee has a name, age, and birthday method as well as a salary and a pay method. Employee

Characteristics of Objects – Polymorphism
Definition Same operation may behave differently for different classes. In simple words, “ One name multiple form” Here operation mean – it’s a procedure or transformation that an object perform or is subject to. For example , Class name is POLYGON Attributes - vertices, border color, fill color. Operations – Draw, erase, fill An implementation of an operation by a specific class is called Method

OO Development refers to the software life cycle.
What is OO Development? OO Development refers to the software life cycle. i.e. Planning, Analysis, Design & Implementation Why OO Development? In essence of OO development is the identification & organization of application concepts, rather than in a programming language.

OO Dev – Modeling Concept, Not implementation
Earlier, OO community focused on implementation part rather than analysis and design. It focuses excessively on implementation mechanisms rather than the underlying thought process that support. An OO development approach encourage software developers to work & thinks in terms of the application throughout software life cycle.

OO development is a conceptual process independent of a programming language until the final stage. OO development is fundamentally a way of thinking & not a programming technique. It can serve as a medium for specification, analysis, documentation & interfacing as well as for programming

OO Development & Graphical notation represents OO concept. OO process consists of building a model of an application & then adding details to it during design. Same notation is used from Analysis  Design  Implementation. So information is not lost or translated into the next stage [Reusability].

System Conception Analysis System Design Class Design Implementation
OO Development Stages System Conception Analysis System Design Class Design Implementation

System Conception means origin of the system.
S/W development begins with - business analyst or users conceiving an application & formulating tentative requirement.

OO Development Stages System Conception Analysis System Design Class Design Implementation

Task of Analyst Analysis
Must work with the requester (client) to understand the problem, because problem statement are rarely complete or correct. To design the Analysis model which demonstrates what the desired system must do, not how it will be done. Analyst is not concerned about implementation decision.

Analysis Analysis model Application model Domain model Description of
real-world objects reflected Within the system Description of the parts of application system itself that are visible to the user. Ex. Bank account is domain model. Application model includes Saving accounts, current account, demat account etc.

OO Development Stages System Conception Analysis System Design Class Design Implementation 295

Task of system designer
- must decide what performance characteristics to optimize. - choose strategy to attack the problem. - making tentative resource allocation. Ex. Designer might decide to change the window screen for fast & smooth working, even when windows are moved or erased.

- add details to analysis model
Class Design Task of class designer - add details to analysis model - They determine data structures & algorithm for each of the operation of window class. - They elaborate both domain & application objects using same OO concept & notation.

Task of Implementers :-
Implementation Task of Implementers :- Translates the classes & relationships developed during class design into particular programming language, database or hardware During implementation, follow good software engineering practice so that traceability to the design is apparent (i.e. clear).

Summary of OO Development
OO concepts apply throughout the System Development Life Cycle (SDLC) i.e. Analysis  design  implementation Use same classes from stage to stage without a change of notation. Some classes are not part of analysis but are introduced during design or implementation. Ex. Data structure such trees, hash table & linked list are not visible to users at the time of analysis. but designers introduce them to support particular algorithms.

Three Software Models Models Class Model State model Interaction model
Objects in the system and their relationship Life history of objects in the system Interaction among objects

Class model classes Book Store Relationship
Describe the static structure of the objects in a system & their relationship It define the context for software development. Class model contains class diagram to express it. A class diagram is graph phase. Nodes are classes. Arcs are relationship among classes. classes Book Store Relationship

State Model Describe aspect of an object that change over time.
State model specifies & implement control with state diagram A state diagram is a graph whose Nodes are state Arcs are transitions between state caused by events State 1 State 2 State Diagram Event

Interaction model Sequence Diagram Elaborate Use Cases
How the objects in a system co-operate to achieve broader results. Interaction model start with Use case that are elaborate into with sequence and activity diagram. Use Cases Activity Diagram Sequence Diagram Elaborate

Interaction Model Use cases: Sequence diagram Activity diagram
Focus on functioning of system Simple mean, what a system does for users Sequence diagram Shows the object that interact Time sequence of their interactions Activity diagram Elaborate important processing steps. Activity diagrams can be used to describe the business and operational step-by-step workflows of components in a system

Are the man behind Invention of OO Modeling Technique.
OO Modeling History Grady Booch James Rumbaugh Ivar Jacobson Are the man behind Invention of OO Modeling Technique. Object Modeling Technique (OMT) concept evolved in

OO modeling History In 1994 James Rumbaugh joined Rational (now the part of IBM) in 1994 & began working with Grady Booch on UML Notations. In 1995, Ivar Jacobson also joined Rational & added his concept to the unification work.

OO modeling History In 1996 the Object Management Group issued a request for proposals for standard OO modeling notation. Later Rational led the final proposal team, with Booch, Rumbaugh & Jacobson deeply involved.

OO modeling History

OO THEME OO themes are not unique to OO systems, they are particularly well supported. Abstraction Encapsulation Combining data and behavior Sharing Emphasis on the essence of an object Synergy

Abstraction: OO THEME Use of Abstraction:
Focus on essential aspects of an application while ignoring details. i.e. focusing on what an object is and does, before deciding how to implement it. Use of Abstraction: Freedom to make decisions as long as possible by avoiding premature commitment to details. Ability to abstract is probably the most important skill required for OO development.

OO THEME Encapsulations ( Information Hiding):
It is separates the external ( accessible to objects) aspects of an objects from the internal ( hidden from other objects) implementation details. It prevents portions of a program from becoming so interdependent that a small change has massive ripple effects. For ex. You may want to change the objects to Improve performance, Fix a bug, Consolidate code, Support porting

OO THEME Encapsulation is not unique to OO language but ability to combine data structure & behavior in a single entity makes encapsulation cleaner & powerful.

Combining data & Behavior
OO THEME Combining data & Behavior In non-OO code, to display the content of a window must distinguish the type of each figure, such as circle, polygon etc & call the appropriate procedure to display it. In OO code, program invoke the “Draw” operations on each figure and each object implicitly decide which procedure to use, based on its class. So caller of an operation need not consider how many implementation exist.

Data structure Hierarchy
OO THEME So, maintenance is easier, because the calling code need not be modified when a new class is added. In an OO system, the data structure hierarchy matches the operation inheritance hierarchy. Data structure Hierarchy Procedure Hierarchy Class Hierarchy Is replaced by

Sharing OO THEME OO technique promote sharing at different levels.
Sharing via inheritance is one of the main advantage of OO language. OO development not only lets you share information within an application, but also offers the aspects of reusing designs & code on future projects. OO provides the tools to build libraries ( or collection) of reusable component.

Emphasis on the essence of an object:
OO THEME Emphasis on the essence of an object: In OO technology, focus is on what an objects is rather than how it is used. Use of an object depend on the details of application and often change during development. OO development greater emphasis on data structure & lesser emphasis on procedure structure.

OO THEME Synergy : OO concepts can be used in isolation but together they complement each other synergistically.

Chapter 14 System Design Prepared By: Ashwin Raiyani.

Similar presentations

Presentation on theme: "Chapter 14 System Design Prepared By: Ashwin Raiyani."— Presentation transcript:

Similar presentations

About project

Feedback

Log in

Auth with social network:

Chapter 14 System Design Prepared By: Ashwin Raiyani.

Similar presentations

Presentation on theme: "Chapter 14 System Design Prepared By: Ashwin Raiyani."— Presentation transcript:

Similar presentations

About project

Feedback