Building a Cisco Network for Windows 2000 P2

Chia sẻ: Thach Sau | Ngày: | Loại File: PDF | Số trang:20

lượt xem

Building a Cisco Network for Windows 2000 P2

Mô tả tài liệu
  Download Vui lòng tải xuống để xem tài liệu đầy đủ

DEN is a solution to several challenges from which both enterprise administrators and software vendors suffer. Administrators and vendors are faced with the following issues:

Chủ đề:

Nội dung Text: Building a Cisco Network for Windows 2000 P2

  1. 4 Chapter 1 • Developing a Windows 2000 and Cisco Internetwork The DEN Solution DEN is a solution to several challenges from which both enterprise admin- istrators and software vendors suffer. Administrators and vendors are faced with the following issues: s How to integrate new e-business systems s How to incorporate service level agreements for specific users s How to apply and manage policies s How to integrate management “islands” (i.e., separate network administration units and separate network management systems) s How to get interoperability from systems right out of the box s How to achieve advanced services that are applicable network-wide DEN solves these issues with the definition of a directory service, shown in Figure 1.2, which can manage: s Integration of e-business systems, media, devices, and protocols s Incorporation of service levels into the management of users and applications s Application and management of policies s Integration of extensible management applications into the direc- tory to centralize the network management s Utilization of common protocols, common application programming interfaces (APIs), and a common repository for information to ensure interoperability s Advanced services from configuration, access control, security, and provisioning of Quality of Service (QoS) As a result, DEN harnesses the power of a database to centralize and manage network systems and services. DEN defines a common schema for network units and services, and enables interoperability between them. DEN specifies an object-oriented information model, called a directory, for networked units. A networked unit is defined within the directory as a class. The network units, or classes, are not limited to devices or user accounts, but encompass every possible application or system that can participate on the network. Classes are composed of objects that share the same basis of attributes. Any single network element (a user account, server, policy, etc.) represents some individual entity (Joe User, Server1, or SecurityPolicyA, and so on) on the network. Each object contains a set of
  2. Developing a Windows 2000 and Cisco Internetwork • Chapter 1 5 Figure 1.2 Directory-enabled networking architecture. Directory Report Report can be generated from directory Application C Application A with integrated information Directory service Distributed storage Application B Application D Users can access directory for use of applications attributes that describe its properties. For example, an attribute for a user account may be the user’s telephone number. DEN does not define a management protocol like Simple Network Management Protocol (SNMP), even though it enables network manage- ment at a new level. It does not define a network protocol like Lightweight Directory Access Protocol (LDAP), although new directory services will likely integrate LDAP. It does not define a new type of schema for a database. DEN is not a product in and by itself. DEN is a definition of the foundational elements required for building a directory enabled network service or application. It defines a standard hier- archy for a directory service, but opposes limitations by defining extensi- bility. When DEN is used, multiple vendors will not experience conflicts between their schemas, and network device configuration and management can be performed through the use of the directory service. In the DEN policy server model, network devices will use standard pro- tocols to access the network, such as Domain Name System (DNS) and Dynamic Host Configuration Protocol (DHCP). The network devices will access servers or hosts to attempt a network transaction, which will check the directory service (whether it is stored locally, or on other servers) for any policies that may apply. If a policy does apply to the network transaction, the policy is applied and the transaction is permitted with whatever alterations the policy requires, or denied based on the policy, as shown in Figure 1.3.
  3. 6 Chapter 1 • Developing a Windows 2000 and Cisco Internetwork Figure 1.3 Policy server model. Policy Policy management Management application Directory Service Stored Policy Policy Server Yes Yes Policy Decision? No Policy Enforcement Network Traffic Flow QoS is a way of establishing a priority (or lack of priority) for a specific type of traffic depending on when it is sent, what type of traffic it is, where it is going, or from where it is coming. Look at an example where it is assumed that a corporate executive videoconferences with direct reports over the internetwork on a monthly basis. This executive travels from one location to another and can be any- where when he or she holds the videoconference. As a result, the executive is never using the same computer or the same Internet Protocol (IP) address when videoconferencing. Many QoS products will mark a type of traffic with priority based on its physical or Media Access Control (MAC) address, which is determined from either the IP address or host name of the computer using Address Resolution Protocol (ARP). If the executive wants the videoconference to be granted priority over other network ser- vices, then the network administrator will need to know what IP address or host name the executive is using at the time the videoconference is held. Not only that, but the administrator will need to find out that same infor- mation each and every time the executive holds a videoconference. Without a network administrator manually configuring the videoconference to have priority through QoS, the videoconference will suffer, and as a result, this type of QoS usage will result in an excessive amount of administrative
  4. Developing a Windows 2000 and Cisco Internetwork • Chapter 1 7 overhead. If the executive holds a spontaneous videoconference without notifying the administrator, then he or she will not receive the expected performance and will be disappointed that the business objective was not met by the QoS product. All of this is a recipe for failure. The type of network environment in which a QoS product using IP addresses for policy definition will work well is a static environment in which the IP addresses, host names, and traffic types rarely change. With the rate of change of technology today, this type of network is rare. A DEN-based QoS product can resolve this issue. A DEN-based QoS product potentially can attach a user’s account dynamically to his or her computer’s IP address at logon, and statically attach the QoS policy to the user’s account. Going back to our videoconferencing executive, he or she would log on to the network and would already have a VideoConference QoS policy attached to his or her user account (the policy having been cre- ated by the administrator and assigned to the user account). At logon, this policy would dynamically be assigned to the IP address the executive had at that moment. The administrator never needs to be involved except for the initial definition of the QoS policy, and the executive always receives the QoS needed for his or her videoconferences, regardless of where he or she logs on to the network. TIP Whitepapers and other information about QoS and policy-based net- working can be found on the Internet at the following addresses: tsld001.htm
  5. 8 Chapter 1 • Developing a Windows 2000 and Cisco Internetwork About Microsoft’s Windows 2000 and Cisco’s IOS Microsoft’s Windows 2000 and Cisco’s Internetwork Operating System (IOS) combine to provide the power of a DEN model. These operating sys- tems are described briefly in the following section, and in much more detail in Chapter 2, “A Tour of Windows 2000,” and Chapter 3, “Cisco Hardware and IOS Basics.” Cisco’s IOS and Software Products Cisco develops a great deal of software products to work with their hard- ware products. The Cisco IOS is a platform that provides network services to an internetwork. It supports both local area network (LAN) and wide area network (WAN) environments, although actual configuration for an environment must also be supported by the Cisco hardware. The IOS can scale to multiple interfaces on a single piece of hardware, and with mul- tiple routers in an internetwork, the IOS proves to be versatile in addition to being scalable from small offices to large enterprise internetworks. IOS supports standard network protocol stacks and media types, including (but not nearly limited to): s Transmission Control Protocol/Internet Protocol (TCP/IP) s Internetwork Packet Exchange/Sequenced Packet Exchange (IPX/SPX) s AppleTalk s Ethernet s Token Ring s Frame Relay s Integrated Services Digital Network (ISDN) s Asynchronous Transfer Mode (ATM) Cisco’s IOS is the operating system that Cisco routers, switches, and access servers use to boot up. To enhance access services, routing, and bridging, the IOS supports a full set of security features—encryption, authentication, access control, packet filtering, and firewall services. The IOS is upgradeable as Cisco releases new versions. Each version includes new capabilities and network services. These new services meet enter- prises’ business requirements for new technology. The IOS can support and grow with an organization’s needs.
  6. Developing a Windows 2000 and Cisco Internetwork • Chapter 1 9 In the grand tradition of UNIX enthusiasts, the IOS is command-line friendly. Although Cisco routers do not come equipped with monitors, they can be accessed over the network, or through a terminal connection. The Command Line Interface (CLI) appears as a simple text-based screen with a prompt, somewhat similar to a DOS prompt. Newer versions of the IOS can be configured using HTML pages and a Web browser. Cisco ConfigMaker Designing an internetwork is not an easy job. It takes knowledge of proto- cols, hardware, software capabilities, and how to place and configure them to achieve the optimal s Performance s Reliability s Availability s Security s Scalability s Manageability These must meet the client’s business requirements, and some are in conflict with others. For example, a highly secure internetwork placed in an environment where usability of the network is the highest priority for a business requirement may not be easily achieved. To the organization, usability may mean granting users short passwords that are identical from system to system and that never change, whereas a highly secure network would absolutely require lengthy passwords that change on a frequent basis. A designer must be aware of these types of issues and be prepared to make decisions based on business requirements. The network designer should make recommendations that are sensible for the environment, even if the organization might want something a little different. In the security versus usability requirements, for example, the network designer could recommend using DEN-compliant systems where all user account informa- tion was held in a single database for the entire internetwork, thus requiring users to need only a single password. Then again, the designer could recommend that the users are trained on having longer passwords using numbers and characters (rather than alphabet-only), and suggest that a policy be put in place to force the users to change the passwords on a 60- or 90-day basis. This may not be the most usable system, but it is a fair compromise! Cisco provides a free tool (yes, FREE!) called Cisco ConfigMaker that a network designer can use when designing an internetwork. Cisco
  7. 10 Chapter 1 • Developing a Windows 2000 and Cisco Internetwork ConfigMaker which is an application that runs on Windows 95, Windows 98, Windows NT, or Windows 2000 (on Windows 2000, you should install the Windows NT version). ConfigMaker is downloadable from, and is shown in Figure 1.4. ConfigMaker is straightforward, allowing the network designer to con- figure a small- to medium-size network, or begin the basic design for an enterprise wide area network, or a section of a large network that does not utilize the enterprise 7x00 series routers that are not listed within the ConfigMaker tool. Each new version adds new equipment and features, but the latest version 2.4 supports Cisco routers from the 800 through the 4000 series, switches, hubs, voice equipment, modems, ISDN, and other network devices. Figure 1.4 Cisco ConfigMaker. Even though the ConfigMaker tool looks similar to other design applica- tions in which you simply drag a network component to the design window and create the connections, it has a couple of additional features. ConfigMaker forces the designer to make critical design decisions while building the design. It will not allow a connection to be created between two routers if either does not have a port available for that connection. It
  8. Developing a Windows 2000 and Cisco Internetwork • Chapter 1 11 Figure 1.5 ConfigMaker router slot configuration. requires you to state the IP addresses of the interfaces, and warns you if you have selected an IP address that is assigned to another network seg- ment. It forces you to apply passwords to the routing equipment. A typical router configuration dialog, illustrated in Figure 1.5, shows how ConfigMaker includes the interfaces available for the slots in a router (in drop-down boxes) so that you can select each interface as you build the router, and do not accidentally select an interface that is not available for that particular device. ConfigMaker can also collect information about a Cisco device on your network, read which interfaces are installed within it, and then put that information into your network design. In addition, ConfigMaker can write configuration files to routers. It can greatly reduce the time and effort it takes to diagram an existing internetwork. The AutoDetect Device Wizard is shown in Figure 1.6. Cisco FastStep Cisco provides another tool, also for use on Windows 95, 98, and NT (or 2000), for configuration of Cisco series 700, 800, 1600 routers and dialup 2500 series access servers. It is called FastStep. This tool is available as a free download at
  9. 12 Chapter 1 • Developing a Windows 2000 and Cisco Internetwork Figure 1.6 AutoDetect Device Wizard can assist in diagramming a network. Figure 1.7 Fast Step for 800 series routers.
  10. Developing a Windows 2000 and Cisco Internetwork • Chapter 1 13 Figure 1.8 Router and option selection in Fast Step. The FastStep application guides an administrator using a “wizard-like” sequence of dialog boxes, such as those shown in Figure 1.7. Each dialog box adds new information towards building a configuration and then applying it to a router. FastStep allows the administrator to select the specific router model and the types of options that the IOS should support on the router. The dialog that illustrates this selection is shown in Figure 1.8. When you complete a router configuration with FastStep, the applica- tion will save your options in a file. This file will be the name you give your router concatenated with the suffix .cfg. So, if you run the FastStep appli- cation and name your router MyRouter, the file will be called MyRouter.cfg. A sample of this file is available in Appendix A. CiscoWorks 2000 Once a router or switch is up and running, the administrator’s next task is to manage it. A network consisting of only one or two routers or switches that is used only during standard business hours (Monday through Friday, 8:00 AM to 5:00 PM) is a simple system and fairly easy to manage on a manual basis. However, if you have a complex internetwork, or one that
  11. 14 Chapter 1 • Developing a Windows 2000 and Cisco Internetwork must be online and available 24 hours a day, seven days a week, then you need to look at the manageability features and management applications. CiscoWorks is one such application for network management. Cisco recommends that CiscoWorks be used in small to medium networks. CiscoWorks is available for use on UNIX, and there is a version available for Windows. There are several components within the CiscoWorks application: CiscoView A graphical view of back and front panels of the Cisco devices, provided remotely in order to simplify monitoring and configuring devices. Show Commands A translator of the IOS command-line language to dis- played router system and protocol data, facilitating a novice adminis- trator’s ability to understand how the router is configured and working StackMaker An application that an administrator can use to create a vir- tual stack of devices for easier visual management Threshold Manager A remote monitoring (RMON) application that can set thresholds on Cisco routers and switches in order to alert an administrator when the device is not working at optimal levels WhatsUp Gold An application licensed from Ipswitch, Inc. that delivers the ability to discover the internetwork’s configuration and to map it, as well as the ability to monitor and track alarms. NOTE More about the Cisco IOS and other Cisco products can be found in Chapter 3. Microsoft’s Windows 2000 Microsoft developed Windows 2000 in answer to a number of challenges. Some of the challenges were from the latest business requirements for Microsoft customers. Other challenges were from technology drivers of the Information Technology industry. And finally, some challenges were posed by the previous Microsoft Windows NT technology—feedback from cus- tomers and critics exhorted Microsoft to transform some of the less usable features of Windows NT. Business drivers are always changing. Organizations need to be com- petitive in order to be of value, and as a result, they are always pushing
  12. Developing a Windows 2000 and Cisco Internetwork • Chapter 1 15 the envelope of technology to assist them in gaining competitive advan- tages. Increasingly, business is being done on the Internet, so much so that it is called e-business. Organizations no longer use their Web site as solely a marketing message, but they also utilize it to: s Enhance their customer relationships by enabling customers to self-manage their data in Customer Relationship Management (CRM) applications s Sell products and services through e-commerce applications s Demonstrate products through remote control applications that run over IP s Streamline accounting, logistics, and other processes between their own organization, vendors, and clients using business-to-business (B2B) applications The Internet delivers a critical requirement for security to businesses. The Internet is the world’s most publicly accessible data interchange. Any network resource or service that is attached to it, any user that sends data across it, is subject to the possibility of being endangered by someone (who for reasons of malice, idleness, challenge, or pure insanity, decided to attack that resource, service, or data). Depending on the use of the resource, service, or data, there are various levels of security that a busi- ness may wish to apply. A network administrator may wish to protect a Web site page used for marketing the company from being altered while it sits on the Web server, but that network administrator would probably not need to encrypt that page while it travels across the Internet. It’s mar- keting, why hide it? However, when creating a B2B system between a vendor and the organization for the streamlining of accounting informa- tion, the network administrator would probably want the data to be encrypted while it crosses the Internet. And for servers that store develop- ment data for a small group of internal developers, the network adminis- trator may not wish that server to be available on the Internet at all. Another challenge that businesses face is the increasing need for mobility in end-users. These users must be able to move from one location to another, and to carry their data and the ability to work along with them. Not only do users move about from one office to another within large enter- prises, but there is an escalating requirement for businesses to support telecommuting. Laptops, once a small or nonexistent percentage of equip- ment attached to a network, are rising to become greater than 30 percent of networked clients in many organizations. Technology is constantly changing. New developments that are becoming more accepted by businesses include:
  13. 16 Chapter 1 • Developing a Windows 2000 and Cisco Internetwork s Data/voice/video convergence s Quality of Service s Virtual private networking (VPN) in the forms of IP Security (IPSec) and Layer 2 Tunneling Protocol (L2TP) s Digital Subscriber Line (DSL) s Policy-based networking Not only does an operating system need to look forward at new tech- nologies, being able to adapt to them quickly, but it also needs to be able to work within an existing infrastructure. New media and old media, new protocols and old protocols, new hardware and old hardware must all be supported. Critics of Microsoft’s Windows NT bemoaned its lack of reliability. They created an acronym, BSOD, for the Blue Screen of Death, describing the blue screen with a nearly indecipherable error message for the problem that caused the NT machine to fail. BSODs were common. What was just as bad, however, was the need to reboot … all the time. Whenever you made a change to the NT machine, especially involving a change to the system itself or the addition of a new application, you were forced to reboot it. If it was a server you were configuring, you then needed to ask all the users to log out of the server while it rebooted so that any files that they were using would not be lost or damaged due to a sudden disconnection. For those organizations who were familiar with Windows NT’s need for rebooting, they created a policy that configuration changes could be com- pleted only after business hours, or during slow periods (if they had extended business hours). In addition to reliability issues, Windows NT was simply not user- or laptop-friendly. It did not detect hardware—instead the hardware had to be configured with special Windows NT drivers. Installing the Windows NT operating system seemed easy enough, unless some component did not have an NT driver. It did not easily connect to the Internet through its remote access client application. If an NT File System (NTFS) partition somehow became corrupted, or a system file within an NTFS partition became corrupted, there was no way (without a third-party utility) to access the NTFS partition and make those system file replacements. The power management capabilities (absolute requirements on laptops) were nonexistent, again depending on the vendor of the hardware to provide any power utilities. Users became frustrated when faced with these challenges. Microsoft created Windows 2000 with solutions to these types of chal- lenges. They made absolutely certain that the new Windows 2000 oper- ating system included the following:
  14. Developing a Windows 2000 and Cisco Internetwork • Chapter 1 17 s Support for e-business through the incorporation of Internet ser- vices, World Wide Web (WWW), File Transfer Protocol (FTP), Network News Transport Protocol (NNTP), and more. (In contrast to Windows 2000’s standard inclusion of these features, Microsoft made these features available in Windows NT v4.0 as an Option pack; the standard operating system did not include them.) s Security mechanisms and new technologies such as QoS, IPSec, L2TP, and DSL. s Data/voice/video, using standards-based protocols for its Telephony Application Programming Interface (TAPI) version 3.0 Call Control and Media Services, and support for multicast for- warding. s Policy-based networking capabilities within the new Active Directory services. s Answers to users’ issues with reliability, usability, and laptop- friendliness. There are very few reasons why a Windows 2000 machine requires rebooting. The BSOD is much more rare than it was in Windows NT. Installation is uncomplicated, since the Windows 2000 will detect hardware and install Plug and Play stan- dard hardware. The user interface (UI) is simplified and the oper- ating system supports power management standards. Active Directory The Windows 2000 Active Directory is a tremendous enhancement to Windows 2000. This is a feature that can be installed only as part of the Windows 2000 Server line. There are four products: Windows 2000 Professional Meant to be installed on workstations and PCs for end-user usage, and is considered the upgrade for Windows NT Workstation v4.0. Windows 2000 Server Intended for small or workgroup servers, and is considered the upgrade version of Windows NT Server v4.0. Windows 2000 Advanced Server Intended for enterprise servers through support for additional processors and network load balancing clusters, and is considered the upgrade for Windows NT Server v4.0 Enterprise Edition. Windows 2000 DataCenter Server ntended for the highest-end servers, supporting up to 32 processors, and released only through Original Equipment Manufacturers (OEMs), it is customized for that manufacturer’s high-end server equipment. This is not considered an upgrade for any Windows NT version.
  15. 18 Chapter 1 • Developing a Windows 2000 and Cisco Internetwork Figure 1.9 Active Directory policy-based networking. Policy Server Directory applies policy to Joe's request and allows access to HPPrinter Policy is applied to HPPrinter, but no access required for Server Checks directory for access to HPPrinter HPPrinter User Joe Active Directory Domain Three products, Windows 2000 Server, Windows 2000 Advanced Server, and Windows 2000 DataCenter Server, can become Active Directory domain controllers (DC). A DC is a server that stores a replica of an Active Directory partition on its storage system. The Active Directory itself is a database, using a structured schema of classes, objects, and attributes (just like the DEN specification) that store data about user accounts, net- work services or applications, and network resources or equipment. In addition, the Active Directory stores policies about each of these compo- nents such that relationships between the objects (individual users, ser- vices, and resources) can be managed. For example, a policy can be created that states a User named Joe can access administrative level com- ponents of a resource named HPPrinter, thus enabling Joe to manage the print jobs that are sent to that printer by other users, but not requiring him to have administrative access to the server that provides HPPrinter services to the network, as shown in Figure 1.9. The Active Directory incorporates all of the features shown in Table 1.1.
  16. Developing a Windows 2000 and Cisco Internetwork • Chapter 1 19 Table 1.1 Active Directory Features Feature Description Active Directory The Active Directory Services Interface (ADSI) is an API for Services Interface programming and scripting Active Directory applications. (ADSI) Domain Name The Active Directory is integrated with DNS, using it as System (DNS) the mechanism for locating services on the internetwork. integration Extensible The Active Directory schema can be extended with schema new objects, classes, and attributes to support new technologies and capabilities. Group Policy Centralized policy management is provided through Group Policy integrated with the Active Directory. Group Policy provides predefined policies for customizing network activity. Hierarchical The Active Directory database is organized into a tree, or architecture hierarchy. Each single Active Directory domain can be subdivided through the creation of Organizational Units (OUs). An OU can contain other objects, including other OUs, thus resulting in a structure much like a file system’s directory structure. Hierarchical Each Active Directory domain is created as its own DNS namespace namespace, and is connected through hierarchy of DNS naming. For example, if were the root domain, a subdomain could be named Kerberos The Active Directory uses Kerberos security as its authentication method. LDAP The Active Directory is compliant with Lightweight integration Directory Access Protocol (LDAP) version 3, so that it will be able to interoperate with other LDAP-compliant appli- cations and services. Multimaster The Active Directory is capable of storing multiple replicas replication of the same partition of its database on different servers. To ensure synchronization between all the replicas, each partition is considered a master and uses an algorithm to resolve any conflicts when the updates to the database are replicated between servers. Scalability Because the Active Directory uses the hierarchy of domains, it can support multiple partitions and can scale up to millions of users, resources, and services.
  17. 20 Chapter 1 • Developing a Windows 2000 and Cisco Internetwork NOTE More about Microsoft Windows 2000 can be found in Chapter 2. Merging together with Cisco Networking Services for Active Directory The Microsoft and Cisco partnership, and their DEN initiative, has resulted in Cisco Networking Services for Active Directory (CNS/AD). CNS/AD extends the schema of the Active Directory to include new object classes for Cisco equipment, as well as new attributes for existing Active Directory objects. By extending the schema with this software, policies can be applied to users from within the Active Directory itself. The simple Windows 2000 user interface enables a point-and-click system for defining some complex administrative tasks. When CNS/AD is installed, new objects exist for the administrator. Routers can be added using their specific version object. Users can also be added as user accounts. Policies can be added to the Active Directory to guide how the user’s traffic will interact with the router. When the router comes online, it recognizes an Active Directory domain controller as its policy server storage. At the time that traffic comes from a source, the router checks the Active Directory for the policy. If the policy includes instructions on handling the traffic, the router follows those instructions. Best Practices for Implementing a Network When you are ready to start planning your Cisco and Microsoft internet- work, you can use the following best practices to ensure a swift, undisrup- tive deployment. The first phase of your deployment will be to gather information about the existing network. Most enterprises already have a network of some type. You will need to discover and assess: s Cabling media s Existing servers, peripherals, and desktop hardware s Operating systems being used s Applications in use
  18. Developing a Windows 2000 and Cisco Internetwork • Chapter 1 21 s Protocols being deployed s Integration with mainframe and other network resources s Security requirements and methods currently used The second phase of your deployment will be to designate the teams for your project. Not only should you use technology experts in each team, but also decision makers and business unit leaders. Although the decision makers may have little to offer in the discussion of the technology, they will have far more insight into how that technology must be able to sup- port their existing business processes. The third phase is to create a vision of the future network. This is the goal of your project. In many cases, it is a good idea to go ahead and put the impossible down as a goal when a team member suggests a technology need. Sometimes that so-called “impossible” feat is actually available as a feature of a technology already available. Other times, the requirement that drives that goal can be met by configuration changes or other technology implementation. When you reject the goal outright, you can miss the opportunity to improve the enterprise. When you discuss your vision, make sure to include the following: s Business requirements s Proposed hardware and software s Needs for Internet integration s Configuration requirements s Future changes in media, servers, desktops, or peripherals s Security requirements In most large businesses, you will then need to ensure that you have executive sponsorship and a budget for moving forward with the project. And after that, you will use the vision to guide the internetwork design, which absolutely requires a network designer who is familiar with the pro- tocols, hardware, software, operating systems, and applications. The designer will need to determine what other network components are required, where they should be placed, how they are interrelated, and how they should be configured. Each design decision should be justified, especially when it is one that will affect the budget for your project. The network design should be business-requirement driven, and should include the following elements: s Desktop environment design s Server environment design
  19. 22 Chapter 1 • Developing a Windows 2000 and Cisco Internetwork s Directory service designs (directory services may require several elements to their design) s Hubs, switches, and LAN design s Routers and WAN design s Strategies for supporting mobile users, Internet, IP Addressing, network management, and network security The network designer should be able to leverage the existing infrastruc- ture as much as possible and design the resulting network structure for a flexible administration. The architecture of the new network should elimi- nate any existing single points of failure and resist bringing any new single points of failure into the internetwork. It should also provide a capacity projection for network bandwidth, router storage, router memory, server storage, server memory, and standard desktop storage and memory. Once the teams and executive sponsors accept the design, the project can begin with providing proof of concept. This is done through a lab. Each design element is tested and developed such that the deployment of the design can be performed with the least risk and the most efficiency. This may result in changes to the design. When the lab team is satisfied that the processes and design are ready for production, the project can begin a pilot. The pilot is basically a mini- project performed for a small subdivision of users, services, and resources on the internetwork. If the pilot is unsuccessful, the team should return to the lab and work out the problems. Once the pilot is successful, however, the project can be fully deployed across the rest of the internetwork. Sometimes, this final deployment is best done in phases consisting of multiple pilots and deployments until everything has been completed. All in all, when deploying a project that involves every section within the internetwork, best practices dictate that the enterprise backbone is completed first, the server and secured access portion of the internetwork is completed next, and the desktops and peripherals are completed last. Networking Basics A solid knowledge of networking basics is required for understanding the concepts in this book. If you need a refresher course, just browse through the OSI Protocol Reference Model, Internet History, and IP Networking Primer sections. Otherwise, if you are confident with your networking basics knowledge, just skip to the Case Studies section and then on to the rest of the book!
  20. Developing a Windows 2000 and Cisco Internetwork • Chapter 1 23 For IT Professionals Resolving Network Bottlenecks Network bottlenecks are sometimes encountered during a lab, but most often are encountered after the network has been in use for some time and users have become comfortable in using it, hence causing more internetwork traffic. The typical causes for a network bottleneck are: s Over-utilized servers, either in memory or storage s High-bandwidth utilization on network media, usually only on a portion of the internetwork s Loss of network integrity You can resolve many bottlenecks by following these best practices: s Balance network traffic to a server by using multiple network adapters and network load balancing. s Use media and protocols that enable the highest bandwidth available. If you have Category 5 cabling and are using shared Ethernet 10BaseT, consider changing to a switched Ethernet configuration or Ethernet 100BaseT, or even both! s Whenever possible, use interfaces that can offload functions from the machine’s CPU, such as IPSec offloading or checksum offloading. s Reduce the number of protocols that you are using on your internetwork. s Use routers, bridges, or switches to segment broadcast domains. OSI Protocol Reference Model In the Information Technology industry, you will run into some oddball ways of doing things. The Open System Interconnection (OSI) Protocol ref- erence model represents one of the strange phenomena that persists in this industry—standards follow usage. You would think that a standard would be created and then it would be put to use. However, to meet the demand for new technology capabilities, to gain a competitive edge, and in keeping with the counterculture attitude of Silicon Valley, a lot of tech- nology is put to use first. If a practice works, if it meets a real technology
Đồng bộ tài khoản