Introduction to Information Systems Analysis Systems Design Application Architecture

1
Introduction to InformationSystems
AnalysisSystems DesignApplication Architecture
Modeling

• INFO 503
• Glenn Booker

2
System Design

• System Design (a.k.a. physical design) includes
  the evaluation of alternative solutions, and
  specification of a detailed solution
• Focus is the shift from logical to physical
  design now technology counts!
• Critical to consider off-the-shelf options

3
Design Strategies

• Modern structured design
• Information engineering
• Prototyping
• Object-Oriented Design (OOD)
• Joint Application Development (JAD)
• Rapid Application Development (RAD)

4
Modern Structured Design

• A.k.a. top-down program design and structured
  programming
• Is a process-oriented technique to break a
  program into smaller pieces (modules) by using
  certain rules and guidelines
• Each module is a set of instructions to perform a
  specific function

5
Modern Structured Design

• Modules should be cohesive - perform only one
  function this helps reusability too
• Modules should be loosely coupled - have minimal
  interaction or dependency on each other
• Capture design in a structure chart, p. 474
  (314)
• Not event driven or object-oriented

6
Information Engineering

• Analysis was based on defining applications to
  support the requirements of various business
  areas
• IE does not have its own unique design technique
  tools such as the ERD are the starting point
• Address process issues later

7
Prototyping

• Developing prototypes
• Encourages end user involvement
• Is well suited to iterative development
• Allows hands-on verification of requirements
• Provides quick feedback
• Often speeds several life cycle phases

8
Prototyping

• But prototyping
• Encourages irresponsible code-and-fix life cycle
• Can only augment paper specifications and other
  design techniques
• Omits several design issues storage, control,
  performance

9
Prototyping

• And prototyping
• Can lead to premature design commitment (lock in
  design options)
• Can reduce creativity
• Can lead to feature (scope) creep
• Can result in slower performance

10
Object-Oriented Design

• OOD uses the results of OOA to design objects
  which will implement the system
• Addresses data and process issues at the same
  time
• Can, for example, focus on defining objects, and
  then defining the methods which communicate
  between them, p. 478 (546)

11
Joint Application Development

• JAD can be used for participative development of
  a design, in conjunction with other design
  techniques

12
Rapid Application Development

• RAD blends other design techniques (structured
  design, information engineering, prototyping and
  JAD) to speed system development
• Often based on using specific tools which
  support the process

13
FAST Design Methodology

• Design covers the following FAST phases
• Decision Analysis (Configuration) Phase, from
  Chapter 5 in the 6th edition
• Procurement Phase (not a separate phase, it is
  addressed during System Design phase in the 6th
  edition)
• Physical Design and Integration Phase, this is
  detailed design before construction of the system

14
Decision Analysis (Configuration) Phase

• The decision analysis phase identifies candidate
  configurations, and picks the best
• Activities are
• 5.1 Identify Candidate Solutions
• 5.2 Analyze Candidate Solutions
• 5.3 Compare Candidate Solutions
• 5.4 Update the Project Plan
• 5.5 Recommend a System Solution

15
Identify Candidate Solutions

• This activity identifies the candidate system
  design solutions, based on business needs and
  possible implementation technology
• Ideas may come from system owners, users, and
  others involved in the analysis effort
• Candidates may be based on the technology
  architecture of your organization

16
Identify Candidate Solutions

• Activity consists largely of brainstorming for
  ideas, followed by research to assess their
  characteristics
• Do not judge the solutions - yet
• Output is a table to describe possible solutions,
  p. 221 (322), the Candidate Systems Matrix

17
Analyze Candidate Solutions

• Now assess each candidate solution for
• Technical feasibility (practical and staffable)
• Operational feasibility (effect on users)
• Economic feasibility (is it cost effective?)
• Schedule feasibility (can it be done on time?)
• Need to define hardware, training, and software
  costs for each solution
• Output feasibility assessment, p. 223 (324)

18
Compare Candidate Solutions

• Based on the relative importance of the
  feasibility criteria (i.e. their weight in ),
  score each of the candidate solutions
• Weighting of feasibility criteria often done by
  the system owner
• Throw out infeasible solutions (those which are
  impossible)
• Pick the best solution!

19
Update the Project Plan

• Does this task seem familiar yet?
• Update the project plan to reflect the chosen
  system configuration
• Reevaluate scope and tasks as needed

20
Recommend a System Solution

• Now use the updated project plan to recommend a
  solution to the system owner, with other key
  interested parties present
• Might give recommendations (the system proposal)
  verbally or in writing
• Salesmanship is often helpful to pitch the
  solution
• Hopefully ends with approval to continue

21
Procurement Phase
p. 487 (326)

• The procurement phase is to assess off-the-shelf
  options for supporting system design
• This phase is a blend of two new tasks for the
  Procurement phase, and three revised tasks for
  the Decision Analysis phase
• Task numbering in the 6th edition looks weird
  its intended to replace the specified steps in
  phases 4 and 5, respectively

22
Procurement Phase

• New tasks for the Procurement phase are
• 4.1 Research Technical Criteria and Options
• 4.2 Solicit Proposals from Vendors
• Revised Decision Analysis phase tasks
• 5A.1 Validate Vendor Claims and Performance
• 5A.2 Evaluate and Rank Vendor Proposals
• 5A.3 Award Contract and Debrief Vendors

23
Research Technical Criteria and Options

• Identify specifications for hardware and
  software - functions, features, and critical
  performance needs - based on the system
  requirements
• May be based on internal standards, and research
  of trade journals and magazines
• Use to define technical criteria, product
  options, and potential vendors

24
Solicit Proposals from Vendors

• Based on the technical criteria, prepare a
  request for quotation (RFQ) or arequest for
  proposal (RFP)
• An RFQ is used for a predefined product
• Use an RFP when the product is ill defined
• The RFQ/RFP must define the selection criteria,
  and classify requirements as mandatory,
  essential, or optional

25
Validate Vendor Claims and Performance

• Proposals from prospective vendors must be
  validated to ensure their individual merits
• For major software proposals, a Software
  Capability Evaluation (SCE) may be used
• Validation may be through site visits, contacting
  other clients, and other research
• Results in validated or invalidated proposals
• Invalidated means you dont believe them

26
Evaluate and Rank Vendor Proposals

• Throw out all invalidated proposals
• Validated proposals can now be evaluated and
  ranked, using the predefined criteria
• This is a form of feasibility assessment, but it
  might include criteria about the stability of the
  vendor, or their level of support
• Results in a recommendation to use a particular
  vendor (we hope)

27
Award Contract and Debrief Vendors

• The vendor selection is approved by management or
  the system owner
• Contracts are prepared, negotiated, and reviewed
  by winning vendor and system owner
• All candidate vendors are briefed on the results
  of the evaluation (if youre nice)
• Update affected interface requirements

28
Physical Design Integration Phase

• This phase bridges the gulf between the user
  requirements and the builders input needs
• 6.1 Design the Application Architecture
• 6.2 Design the System Database
• 6.3 Design the System Interface
• 6.4 Package Design Specifications
• 6.5 Update Project Plan

29
Design the Application Architecture

• Defines technologies used by the system,
  including data, process, interface, and network
  components
• How will data, process, and interface be
  distributed across locations?
• Based on data and process models
• Might include physical data flow diagram, p. 482
  (373)

30
Design the System Database

• System database(s) are designed in more detail
  allowing not only for full 3NF, but also
  addressing performance, design flexibility,
  storage, security, record size, and
  backup/recovery needs
• Capture in a database schema (often ERD)

31
Design the System Interface

• Design the user interfaces to the system inputs
  and outputs
• Design preprinted forms and reports
• Lots of existing user input is helpful!
• Consider ease of learning, as well as use
• Define controls to help get complete and correct
  data (minimize input errors)

32
Package Design Specifications

• Take the results of the previous three steps and
  put it into a design specification document
  essentially a blueprint to guide the system
  builders
• Need to balance how much design must be specified
  for the builders, and how much can be left to
  their discretion and creativity
• Review with all affected parties

33
Update Project Plan

• Mary had a little lambWhose fleece was white as
  snowAnd everywhere that Mary wentShe updated
  the project plan to reflect her progress
• Update the project plan, now that full design
  information is available (last sanity check
  before the start of construction!)

34
Application Architecture

• Return to general design focus, not detailed
• Application architecture addresses general system
  design and technology issues
• How distributed is computing (processing)?
• How distributed is data storage?
• Make or buy technology?
• Which technologies are used for user and system
  interfaces, and for programming?

35
Modeling Application Architecture

• Application architectures can be captured in
  many ways
• A physical data flow diagram (PDFD) can show the
  technical design, such as the software or people
  involved in each process
• Physical data flows can identify both the name of
  each data flow, and how it is implemented
  (persons role, bar code, HTML, VB, etc.)

36
Modeling Application Architecture
p. 504 (373)

• Physical data flows expand on the logical model
  may also show user vs machine lines
• Physical data flows may include
• Input or output to/from a process
• Database commands (CRUD) with data stores
• Import or export of data from another system
• Flow of data between two modules or subsystems
  within your system

37
Modeling Application Architecture

• Physical data stores might include
• An entire database
• A table in a database
• A file on the computer (whether temporary or
  permanent)
• A tape backup device
• Any other kind of file, paper, or form

38
Modeling Application Architecture

• A network topology data flow diagram can show
  which processors or people support various
  processes
• System flowcharts were popular for showing all
  programs, inputs, outputs, data, and processes
  at once - whew!
• CASE tools can automate generation of flowcharts
  and DFDs

39
IT Architecture

• IT architecture is evolving rapidly stay tuned
  for current events! (see SEI, ACM)
• Networking architecture issues drive many others,
  so look at them first
• Historic footnote early PCs werent designed for
  multiple users or local networks
• Wireless communications emerging as newest key
  factor in networking

40
Five Application Layers

• Presentation layer is what the user sees
• Presentation logic layer determines what needs to
  be displayed on the screen
• Application logic layer controls how the
  application runs
• Data manipulation layer controls getting data,
  which lives in the data layer

41
IT Architecture

• Architecture options range from centralized to
  distributed in varying degrees
• Purely centralized computing was the mainframe
  philosophy (one box does it all)
• Web-based systems are the most decentralized
• Servers are computers shared by many users at
  once
• Clients are used by one user at a time

42
IT Architecture

• Servers might fulfill many purposes
• File servers hold data files used by many users
• Database servers run the database program and
  store its data
• Application servers get data inputs from the
  clients, and get data from the database server
• Web servers host a web site, and communicate
  among clients and application servers

43
IT Architecture

• Clients can be fat or thin
• Fat clients have a custom application installed
  to communicate with the servers (now rare)
• If you had to install a special program to use
  that system, you are probably using a fat client
• Thin clients communicate with the servers through
  an ordinary (not customized) program, such as a
  web browser (now common)
• Thin clients act like a dumb terminal

44
IT Architecture

• Architecture options (see p. 511 (355)) are
• Centralized computing (not shown)
• File server architecture
• Distributed presentation (2 tier)
• Distributed data (2 tier)
• Distributed data application (n tier)
• Network computing (web)

45
Centralized Computing

• Centralized computing was the only architecture
  approach through the 1970s
• Used a mainframe or minicomputer to do all the
  work, accessed through terminals
• VT100 or IBM 3270 terminals really dumb clients,
  they only displayed what text the mainframe sent
  to them
• Very few systems still use this

46
File Server Architecture

• This primitive form of distributed computing
  stores data on a file server, and everything else
  is done by the clients
• The file server just holds the shared data files
  and lets the clients read it or write to it when
  needed
• Is a common cheap way to share data across a
  local area network (LAN)

47
Client/server Architecture

• Client/server architecture (distributed or
  cooperative computing) is the dominant network
  architecture starting in the 1980s
• Data, software, and interfaces may be distributed
  across many clients and servers
• Distributed presentation, distributed data,
  and distributed data and application structures
  are all types of client/server architecture

48
Server Types

• Client/server may use many server types
• Database server for data and data manipulation
• Transaction server to control groups of business
  transactions together
• Application server for the same layer
• Messaging or groupware for Notes or Exchange
• Web server for hosting Internet or Intranet sites

49
Distributed Presentation

• Distributed presentation (a.k.a. chintzy
  client/server) puts a pretty GUI on what were
  text-based (character user interface, or CUI)
  terminal screens
• May use CASE tool called a screen scraper to
  generate a quick GUI from a CUI
• Client does presentation and pres. logic layers
• Still is fundamentally centralized computing

50
Distributed Data

• Distributed data (two-tier client/server) puts
  stored data on the server, and business logic and
  user interfaces on the clients
• A local or wide area network (LAN or WAN)
  connects clients and servers
• Reduces network traffic compared to file server
  architecture easier to maintain data integrity
• This is classic 1980s client/server design

51
Distributed Data and Application

• Also called three-tiered or n-tiered
  client/server, this distributes data and
  application logic to separate servers
• Adds an application or transaction server to
  monitor transactions handle application logic
  functions
• Client still handles presentation and pres. logic
• Is relatively new but increasingly common

52
Network Computing

• Network computing distributes presentation logic
  to web servers, while web browsers handle only
  the presentation layer
• Otherwise looks the same as n-tier client/server
• Is the foundation for most e-commerce systems
• Maintaining security is a major challenge
• Emerging as foundation for widely distributed
  computing and data structures

53
Data Architecture

• Relational databases are the most common tool
  used across distributed data architecture
• The database engine is what handles data
• Data may be distributed across servers
• One table may be partitioned vertically (fields)
  or horizontally (records) across several servers
• Data may be replicated across servers, too

54
Interface Architecture

• Many options exist for handling interfaces
• Batch input and output is used to handle a large
  group of transactions at once
• On-line processing handles transactions as
  needed
• Remote batch processing handles a batch somewhere
  else good for mobile users

55
Interface Architecture

• Keyless data entry tries to eliminate high volume
  input errors, using optical character recognition
  (OCR) or bar codes
• Pen input is often used for simpler data
• Graphical user interfaces are the de facto
  standard for most applications
• E-mail has grown into groupware, for coordinating
  activities

56
Interface Architecture

• Electronic data interchange (EDI) conducts
  transactions among businesses
• Image and document interchange is like EDI but
  captures literal images (photos)
• Middleware helps apps communicate with each
  other, such as ODBC database systems
• These options may be used company-wide, or left
  for individual projects to select

57
Process Architecture

• Software development environments (SDEs) are
  generally selected to match the network
  architecture
• How a system will be deployed affects how it is
  developed
• Centralized and distributed presentation
  typically use COBOL, with a CICS transaction
  monitor, and VSAM or DB2 for managing files and
  data

58
Process Architecture

• Two-tier client/server is the most mature option
  for selecting an SDE
• Tools like PowerBuilder, Delphi, Visual Studio
  are available SQL commands are understood by all
  major database engines
• Testing, debugging, testing, writing, and help
  authoring tools are also available
• Clean layering enforces separate presentation,
  application, and data layers

59
Process Architecture

• Multi-tier client/server typically must support
  1000 users and 1 TB databases
• Client and server must support cross-platform
  environment - Windows, Unix, Mac
• Coding is often object-oriented C, C, Ada95,
  Smalltalk, etc.
• Tools include Forté, IBM VisualAge, etc.

60
Process Architecture

• Network computing uses purely Internet-based
  tools
• HTML for page content and hyperlinks
• Computer Gateway Interface (CGI) components,
  using Perl, Visual Basics ActiveX, etc.
• Java and its cousins
• XML ( SGML lite) to enhance understanding of
  document content

61
Application Architecture Strategy

• Enterprise AA strategy defines corporate
  standards for technology, tools, processes
• Tactical AA strategy defines standards from
  scratch for each project, based on their
  research and needs
• A key part of these choices is whether to buy
  existing tools, or build custom ones

62
Network Topology Overview

• There are four major topologies (structures) for
  a physical computer network
• Bus
• Ring
• Star
• Hierarchical
• Each may use specific protocols (languages) to
  communicate among the computers

63
Network Topology - Bus

• A key function to achieve connectivity among
  computers is definition of the network topology
• The bus is the simplest topology - it connects
  computers along a single line
• A peer-to-peer (serverless) network uses the bus
  topology, with AppleTalk or Ethernet

64
Ring Topology

• The ring topology connects computers in a loop,
  and passes packets of data from one computer to
  the next
• IBMs Token Ring is the most famous example
• Bad if a computer falls off the network hard to
  tell who dropped the packet

65
Star Network

• Many clients may be connected to each server
  across a star network
• The middle of each star is a server
• Servers are connected to each other
• Very common think of using a server for each
  department, and be able to control which
  departments can see each other

66
Hierarchical Network

• This is a set of nested star networks, looks kind
  of like an organization chart
• Generally the servers get smaller the further
  down the hierarchy
• Sometimes used in very large corporations, where
  the biggest servers store company-wide or
  division-wide data

67
Network Protocols

• Networks allow computers to communicate using
  standard languages (protocols)
• Most common are Ethernet and TCP/IP
• Others include Token Ring, IPX, SNA, and
  AppleTalk
• Network operating systems
• Windows 2000 Server 2003 Server, Unix, Linux,
  NetWare, OS/2, Mac OS X Server