PHASE 3: SYSTEMS DESIGN Chapter 8 System Architecture
Chapter Objectives Provide a checklist of issues to consider when selecting a system architecture Describe servers, server-based processing, clients, and client-based processing Explain client/server architecture, including tiers, cost-benefit issues, and performance considerations Describe the impact of the Internet on system architecture 3
Chapter Objectives Explain the difference between online and batch processing Define network topology, and provide examples of hierarchical, star, bus, and ring network models Explain network protocols and licensing issues 3
Chapter Objectives Explain system management tools and techniques, including performance measurement, fault management, backup, and disaster recovery Describe the systems design specification and explain the contents of each section
Introduction An effective system combines elements into an architecture, or design, that is flexible, cost-effective, technically sound, and able to support the information needs of the business System architecture translates the logical design of an information system into a physical structure that includes hardware, software, network support, processing methods, and security 4
System Architecture Checklist A systems analyst must approach system architecture with an overall checklist Enterprise resource planning (ERP) Initial and total cost of ownership (TCO) Scalability Web integration Legacy system interface requirements System security Processing options
System Architecture Checklist Enterprise Resource Planning (ERP) The objective of ERP is to establish a company-wide strategy for using IT resources Describes environment – platform Supply chain management Initial Cost and TCO During the final design stage, you make decisions that will have a major impact on the initial costs and TCO for the new system You should review all previous cost estimates ERP: ERP provides centralized, integrated software applications to help manage and coordinate the ongoing activities of the functional units of an enterprise, including manufacturing and distribution, accounting, finance, sales, product planning, and human resources. In CIO magazine, 5 major reasons companies undertake ERP were cited: Integrate financial information Integrate customer order information Standardize and speed up mfg processes Reduce inventory Standardize HR information ERP also helps companies: Reduce the workforce by eliminating manual tasks and duplication Reduce information system development staff by using packaged solutions Improve customer service through better logistics Improve data integrity through a common database Supply Chain Management: Many companies extend their ERP systems to their suppliers and customers through supply chain management. This can help companies achieve faster response, better customer service, and lower operating costs. RFID allows companies to track inventory items through the supply chain.
System Architecture Checklist Initial Cost and TCO Ask questions like the following Is in-house development still best choice? Is specific package still best choice? New outsourcing available? New technology available? Answers might affect the initial cost and TCO for the proposed system Ask: If in-house development was selected as the best alternative initially, is it still the best choice? If a specific package was chosen initially, is it still the best choice? Have any new types of outsourcing become available? Have any significant technical developments occurred that could affect proposed projects?
System Architecture Checklist Scalability Scalability, also called extensibility Especially important in implementing systems that are volume-related, such as transaction processing systems Web Integration An information system includes application programs, also called applications Web-centric Scalability: Refers to system’s ability to expand, change or downsize to meet changing business needs. Ensures that a system designed, built, or purchased when a company is small is adequate when the company doubles in size. When a system no longer scales, it must be replaced with a new system. Web Integration: A Web-centric architecture follows Internet design protocols and enables a company to integrate the new application into its e-commerce strategy (whether on the internet, co. intranet, or extranet) A web-based application avoids many of the connectivity and compatibility problems that arise when different h/w environments are involved. (external partners use web browsers to import/export data)
System Architecture Checklist Legacy System Interface Requirements The new system might have to interface with one or more legacy systems Interfacing a new system with legacy systems involves analysis of data formats and compatibility The analyst must know if the new application eventually will replace the legacy system Legacy System Interface Requirements: Legacy systems must be considered when evaluating physical design. The ability of a system to share information with other systems sometimes is called interoperability. Systems with interoperability are said to be open, while systems without interoperability are said to be closed. An interoperable system not only can interface with legacy systems, it also is more likely to be able to interface with new systems that are introduced as a result of company acquisitions or mergers.
System Architecture Checklist Processing Options Online or Batch 24/7 Backup and Recovery Processing Options: In planning the architecture, designers also must consider how the system will process data – storage, network, processing resources are different: for instance an online order entry system Vs. a monthly billing system will be much different. Then you must consider will this system be: 24/7 – will system be running 24 hours a day, seven days a week Provision must be made for backup and speedy recovery in the event of system failure
System Architecture Checklist Security Issues Security is a concern at every stage of system development The systems analyst must consider security issues that relate to system design specifications and determine how the company will address them Web-based systems introduce additional security concerns
Planning the Architecture Every information system involves 3 main functions: Data Storage and Access Methods Application Programs Interface The 3 functions may be performed on a … Server OR Client OR Divided between Server and Client Every information system involves three main functions: data storage and access methods application programs to handle the processing logic an interface that allows users to interact with the system Depending on architecture, the 3 functions are performed on a … Server Client Divided between Server and Client You will need to determine where the functions will be carried out and the advantages and disadvantages of each design approach.
Client/Server Architecture Today’s interconnected world requires an information architecture that spans the entire enterprise Whether you are dealing with a departmental network or a multinational corporation, as a systems analyst you will work with a distributed computing strategy called client/server architecture A server, sometimes called a host computer, controls access to the hardware, software, and other resources on a network and provides a centralized storage area for programs, data, and information. Mainframe Architecture – A system design where the server performs all the processing. Centralized Architecture – A server that supports a large number of clients at various locations is called a centralized system. Background: In the 1960s, mainframe architecture was the only design around. Everything was done at a central location. Users had no input or output capability, except for printed reports distributed by corp. IT. As networks advanced and became affordable, companies installed terminals at remote locations, so users could enter and access data from anywhere. This terminal included the keyboard and display screen but no processing capability. (pg. 369) In the 1980s, microcomputer use exploded. Users ran their own Word Processing, Spreadsheet, and Database applications w/o IT assistance in stand-alone computing. Stand-alone computing raised major concerns about data security, integrity, and consistency. Without central storage, stand-alone computing was impossible to protect and back-up.
Client/Server Architecture Overview Client/server architecture File Sharing Architecture – The problem of stand-alones was solved with LANs or FSAs and joining these individual stand-alones or clients. However, this type of architecture requires significant resources since the data file is sent back to the requestor as you can see here. Client/Server: Rapid increases in transmission capacities, significant reductions in cost, and standardized methods of connecting and working with networks have allowed distributed computing to become the preferred method of deploying business applications. Client/server architecture is a popular distributed computing strategy. (Distributed system is one or more connected LANs or WANs. WAN-long distance, continent apart) Client/Server Architecture consists of: The systems that divide processing between one or more networked clients and central server. The clients are other computers on a network that rely on the server for its resources.
Client/Server Architecture Client/Server Design Styles Many forms Database Server Transaction Server Web Server Client/server designs can take many forms, depending on the type of server and the relationship between the server and the clients Servers that perform a specific task, such as those shown in this figure, sometimes are called dedicated servers Database Server – processes individual SQL commands Transaction Server – handles a set of SQL commands Web Server – sends and receives Internet-based communications
Client/Server Architecture Types of Clients: Fat and Thin Fat client - Thick client Thin client Comparison: Thin better performance Thick more processing Fat Client: AKA Thick Client. Locates all or most of the application processing logic at the client Thin Client: Locates all or most of the processing logic at the server Most IT experts agree that thin client designs provide better performance, because program code resides on the server, near the data In contrast, a fat client handles more of the processing and must access and update the data more often Look at pg.374 8-15.
Client/Server Architecture Client/Server Tiers Two-tier design Three-tier design Middle layer Two-tier design – Early client/server designs were called 2 tier designs. Here, the user interface resides on the client, all data resides on the server, and the application logic can run on the server, the client or the client and server. Three-tier design – Recently popular. In a 3 tier design, because the layers are relatively independent of one another, they can be placed on different computer systems, with network connections and middleware servers to bind them into a single application. User interface runs on a client Data stored on server Middle layer between client and server - processes client requests and translates them into commands understood by the server. Think of middle layer as an application server, because it provides the application logic, or business logic Three-tier designs also are called n-tier designs – may use more than 1 intermediate layer. The middle layer is more efficient and cost-effective in large-scale systems because it reduces the data server’s workload and relieves clients of complex processing tasks. Look at pg. 375, fig. 8-17 to compare architectures.
Client/Server Architecture Middleware -Cost-Benefit Issues Scale the system http://www.centene.com/investors/annual_reports Transfer applications Improve response times Client/server systems enable the firm to scale the system in a rapidly changing environment Example: Company like Centene that specializes in adding a new Medicaid plan every year with Client/server computing also allows companies to transfer applications from expensive mainframes to less expensive client platforms Client/server systems reduce network load and improve response times Middleware sometimes is called plumbing because it connects two sides of an application and passes data between them. Common types of middleware include teleprocessing monitors, remote procedure calls (RPCs), object request brokers (ORBs), and distributing computer environment (DCE).
Client/Server Architecture Client/Server Performance Issues Separation of server-based data and networked clients Separates applications and data Client contacts server only when necessary Distributed Database Management System (DDBMS) Although there are many advantages over file-based systems, there can be performance issues of client-server architecture that relate to the separation of server-based data and networked clients In contrast to the centralized system where the command is executed by the server’s own CPU, a client/server design separates applications and data. Therefore, processing is more efficient in centralized because data and program instructions travel together on an internal bus. Client/server systems must be designed so the client contacts the server only when necessary and makes as few trips as possible. Otherwise, bog down. Distributed database management system (DDBMS) Data can be stored in more than one location using DDBMS. Advantages: Data stored closer to users can reduce network traffic New data sites added without reworking system design Data stored in various locations System less likely to experience catastrophic failure Disadvantages: Data security Difficult to maintain controls and standards, when in various locations More complex and difficult to manage
Internet-Based Architecture The Internet has had an enormous impact on system architecture To support this trend, systems analysts must suggest e-commerce strategies that apply available technology and meet their company’s business requirements The Internet has had a huge impact on system architecture. Experts agree that the key to the Internet alternative is flexibility. Implementing e-commerce applications via the Web, an intranet, or an extranet offers: Accessibility — Web browsers and Internet connections are almost everywhere, so applications are available to a large number of users. Low cost — The WANs that form the Internet backbone primarily are government-funded, so connections can be purchased at low cost. Wide availability — Web standards are well-known, so server, client, and application development software is widely available and relatively inexpensive. Potential negative aspects include problems with security, reliability, throughput, and changing standards.
Internet-Based Architecture Developing E-Commerce Solutions In-House If you decide to proceed with an in-house solution, you must have an overall plan to help achieve your goals An in-house solution usually requires a greater initial investment, but provides more flexibility for a company that must adapt quickly in a dynamic e-commerce environment Advantages: Satisfies unique business requirements Minimizes changes in business procedures and policies Meets constraints of existing systems Meets constraints of existing technology Develops internal resources and capabilities Disadvantages: Increases costs Requires more time to implement Lacks proven reliability and performance benchmarks Requires more technical development staff Demands in-house development of upgrades Lacks input from other companies
Internet-Based Architecture Packaged Solutions and E-Commerce Service Providers Turnkey systems Application service provider (ASP) Does lower initial cost outweigh disadvantage of reduced flexibility? Many vendors offer turnkey systems for companies Another alternative is to use an application service provider (ASP) With packaged solution, must consider whether the advantage of lower initial cost outweighs the disadvantage of reduced flexibility later on
Internet-Based Architecture Corporate Portals A portal is an entrance to a multifunction Web site A corporate portal can provide access for customers, employees, suppliers, and the public
Network Models The OSI Reference Model – consists of 7 layers Application layer Presentation layer Session layer Transport layer Network layer Data link layer Physical layer Before looking at Network Topologies, we will briefly discuss the OSI (Open system interconnection) model which describes how data moves from an application on one computer to an application on another networked computer. The OSI model is a logical model that consists of 7 layers. Each layer performs a specific function: Application layer: provides network services requested by local workstation Presentation layer: assures that data is uniformly structured and formatted for network transmission Session layer: defines control structures that manage the communications link between computers Transport layer: provides reliable data flow and error recovery Network layer: defines network addresses and determines how data is routed over the network Data link layer: defines specific methods of transmitting data over the physical layer, such as defining the start and end of a data block Physical layer: contains physical components that carry data, such as cabling and connecters Data goes down through the layers on the transmitting computer, then up through the layers on the receiving computer.
Network Models Network Topology Hierarchical network The way a network is configured is called the networked topology. LANs and WANs are arranged in 4 configurations: Hierarchical Star Bus Ring Concepts are the same regardless of the size of the network, but the implementation is different Hierarchical Network One computer (mainframe usually) controls entire network. Satellite computers or servers control lower levels of processing and network devices. Example: retail chain central computer stores sales activity and inventory levels, local computers handle store operations. Stores transmit data to central computer. Central analyzes stock levels and coordinates supply chain mgmt system. Disadvantage: if business adds additional processing levels, network becomes more complex and expensive to operate and maintain
Network Models Network Topology Star network Star network: Central computer with network devices connect to it. At center is the hub or central computer that manages the network. Star networks are fairly easy to install and maintain. Nodes can be added to or removed from the network with little or no disruption. If one node fails, the other nodes can continue to operate normally. This topology is used when network control is important, because all traffic must flow into and out of the hub. Central computer does not have to serve as primary data storage location; it can serve as network coordinator that enables other devices to transmit and receive stored data from each other. A failure in any star network cable will only take down one computer's network access and not the entire LAN. (If the hub fails, however, the entire network will fail.) Disadvantage: Entire network depends on the central computer. Most large star networks have backup systems in case of a failure.
Network Models Network Topology Bus network Bus network: A single communication path connects the mainframe comuter, server, workstations, and peripheral devices. Information is transmitted in either direction from any workstation to another workstation, and any message can be directed to a specific device. Bus networks are popular on LANs because they are inexpensive and easy to install. Although it can be wired to look like a star network with a wiring hub at the center, a bus network can be identified by the single cable that exists inside the hub. Advantage: Devices can be attached or detached at any point w/o disturbing the rest of the network. Failure in one workstation does not necessarily affect other workstation on the network Disadvantage: Performance can decline as more users and devices are added, because all message traffic must flow along the central bus. If more than a few dozen computers are added to a network bus, performance problems will likely result If more than a few dozen computers are added to a network bus, performance problems will likely result. This does not occur in the treelike flow of a hierarchical network or the hub-and-spoke of a star network where network traffic in one path does not affect traffic elsewhere on the network.
Network Models Network Topology Ring network Ring Network: Resembles a circle of computers that communicate with each other. Although a ring network can span a larger distance than a bus network, it is more difficult to install. A ring network is used when processing is performed at local sites rather than at a central location For example, users accessing computers in the Acctg, Personnel, and Engineering depts perform the processing for individual functions and then use the ring network to exchange data with other computers on the network. Although the ring topology primarily is used for LANs, but it also can be used in WANs. Data flows in only one direction on a ring network. Found in some offices or school buildings Disadvantage: If a network device (Pc or printer) fails, the devices downstream cannot communicate with the network A second difficulty is that if two computers are sending messages at the same time, the messages can become garbled
Network Models Network Topology Other topologies http://compnetworking.about.com/od/networkdesign/a/topologies.htm
Network Models Network Protocols and Licensing Issues The network must use a protocol A popular network protocol is Transmission Control Protocol/Internet Protocol (TCP/IP) A familiar example of a TCP/IP protocol is the file transfer protocol (FTP) Protocols or standards must be used which govern network data transmission. Transmission Control Protocol/Internet Protocol (TCP/IP). Originally developed by US Dept of Defense. Backbone of internet TCP/IP manages the transmission of data by dividing it up into small pieces, or packets. Each packet contains data, as well as the information about the sender, the recipient, and the sequence, which is used to reassemble the data. Example usage: FTP a way to copy files from one computer to another over a TCP/IP network, such as the Internet or an intranet. Consider s/w licensing restrictions in network design since some vendors limit number of users or computers that can access computers simultaneously.
Systems Design Completion System Design Specification Baseline Contents vary depending on company standards You do not have to turn this in System Design Specification: aka technical design spec or detailed design spec presents the complete design for the new information system, along with detailed costs, staffing, and scheduling Baseline against which operational system will be measured Users can understand this document Contents vary depending on company standards and complexity of system
Systems Design Completion User Approval of … Interface design Report and menu designs Data entry screens Source documents Other Users must review and approve the interface design, report and menu designs, data entry screens, source documents, and other areas of the system that affect them Other IT department members also need to review the system design specification When the system design specification is complete, you distribute the document to a target group of users, IT department personnel, and company management
Systems Design Completion Presentations Systems analysts, programmers, technical staff Department managers and users Company management Objective: Obtain management’s approval/support At the end of the systems design phase, you will give presentations: The first presentation is to the systems analysts, programmers, and technical support staff members Your next presentation is to department managers and users from departments affected by the system The final presentation is for company management Key objective: to obtain management’s approval and support for the next development step
Chapter Summary The analyst must consider enterprise resource planning, initial cost and TCO, scalability, Web integration, legacy interface requirements, security, and processing options System security is an important concern An architecture requires servers and clients Compared to file server designs, client/server systems are more scalable and flexible 49
Chapter Summary Any questions? Networks allow the sharing of hardware, software, and data resources in order to reduce expenses and provide more capability to users The way a network is configured is called the network topology The system design specification presents the complete systems design for an information system Any questions? 49