Analyzing the Impact of Infrastructure Design
Analyze the impact of infrastructure design on the existing and planned technical environment.
After thoroughly examining the existing network infrastructure and matching the company's goals to the proposed network infrastructure, you can begin to determine just how drastic the changes will be when moving from the existing network infrastructure to the newly designed network infrastructure. In some cases, you might find that there is a tremendous gap between the existing network and the proposed network. In other cases, the gap will be less extensive. In either case, you will want to minimize the impact on productivity of making the change from old to new. Ultimately, the new network infrastructure design will save the company money and allow it to meet its business goals in the most effective and efficient manner. During the implementation phase, while the network is "under construction," the loss of productivity and increased costs can be significant. A great design includes an analysis of the potential impact of the implementation so that an effective implementation plan can be developed to minimize the costs associated with rolling out the new design. Some factors to consider when determining the impact of implementing your network infrastructure design are described in the following sections.
Perhaps the most important thing to keep in mind when creating a network infrastructure design is the way the customer will actually use the infrastructure once it is in place. Consider the applications that the company will be running. First, list all the applications that are currently in use by the organization. Examine each of the applications to determine its requirements in terms of the network infrastructure. Some applications will be very demanding of the network infrastructure, generating heavy traffic and requiring high throughput, and others will not.
Legacy applications, for example, might not place heavy demands on the network infrastructure with high bandwidth or quality of service requirements, but they might be challenging to support in other ways. Legacy applications might require the use of older, less-efficient protocols. You might not prefer to operate these protocols on your new network infrastructure, but you might find that you must do so in order to support legacy applications that the company refuses to replace or retire.
Once you have compiled a list of all the applications currently in use by the company, you will also want to prioritize these applications. This will help you determine which applications are critical to the company's goals and mission. Critical applications will need to be addressed first when designing the network infrastructure to support application requirements, and then you should address the needs of less-critical applications. If you find conflicts between the requirements of applications, criticality will be one criterion to help determine where to compromise.
Network Infrastructure, Protocols, and Hosts
A computer network is comprised of many parts. There are the individual computer systems that occupy the network, as well as the myriad devices used to connect those systems. This section is all about network infrastructure, which includes all workstations, servers, interconnection devices, and wiring. The most basic level of this network infrastructure consists of the wiring and connecting devices that allow data to travel from one computer to another.
Reference Material Suggestions
As you get into this material, you might want to reference the New Riders Training Guide and Fast Track on the CompTIA Network+ exam for more details. Even though Microsoft eliminated the Networking Essentials exam in the Windows 2000 MCSE Track, it still considers this prerequisite knowledge that can be tested on any exam.
Here are some examples of the types of wiring used in computer networks:
Unshielded Twisted Pair (UTP). This type of wire is also used to connect telephones. Different categories of UTP wire exist, each with its own capacity for carrying data and resisting interference. Categories that you might see in use for computer networks include Category 3 (CAT 3) (even though it's used primarily for voice and very slow networks), Category 5 (CAT 5), and the relatively new Category 6 (CAT 6).
Shielded Twisted Pair (STP). This type of wire is similar to UTP, except that it includes additional protection from external interference (the "shielding"). An example of STP wire is Type 1 cable, used for Token Ring networks.
Coaxial cable. This type of wire consists of a single conductor in the center of the cable, insulated by a protective coating and then wrapped with a stranded or meshed wire. Computer networks use different types of coaxial cable. "Thicknet" is a term that refers to the thick, heavy coaxial cable used by older Ethernet networks. "Thinnet" refers to the relatively newer, lighter, and more flexible thin coaxial cable also used in Ethernet networks.
Fiber-optic cable. This cable type is actually a thin, hollow strand of optical fiber that is used to carry light from one end to the other. Electrical impulses, representing the 1s and 0s of binary data, are converted into pulses of light (on and off) that are carried through these optical fibers. Fiber Distributed Data Interface (FDDI) is a technology designed to use fiber-optic cable.
EIA/TIA 232. These cables, formerly referred to as RS-232 cables, are used to connect serial devices, such as a PC to a modem. These cables might be seen connecting WAN devices.
In addition to the physical wiring, the network infrastructure also includes devices that are the connection points of the network. Such devices include the following:
Repeater. Repeaters are devices that are placed on a computer network to extend the distance that the network may span. As data is carried across a wire, it travels in the form of an electrical signal. This electrical signal loses its intensity as it moves across the wire, so the ability to interpret the data it represents begins to degrade. This is known as signal attenuation. A repeater is used to receive an electrical signal before it degrades completely and repeat it to the wire at full strength. Note that a repeater simply repeats the signal it receives. If the signal is badly distorted when it is received, that distortion is simply amplified when the signal is repeated.
Hubs. A hub is a device that interconnects multiple segments of network wiring. Usually several of those wires connect at the other end to individual PCs. An active hub is actually a kind of repeater. When an active hub receives a signal on one of the wires connected to it, it repeats that signal to every other wire that is connected to it.
Bridge. A bridge is a device that connects multiple physical wiring segments to form a single logical segment. Unlike repeaters, which have no built-in intelligence, bridges can examine the data carried by the wire and make decisions as to where to repeat the signal. These decisions are made based on the MAC address information that the bridge receives with each data frame.
Switch. A switch can be considered a "multiport intelligent bridge." Switches provide tremendous flexibility in deciding where and when to repeat data signals. Some of the latest switches operate at OSI layers 2 and 3 (Data Link and Network) and perform the functions of a multiport intelligent bridge as well as the functions of a multiport router, called a VLAN switch.
Router. A router is a device that makes complex decisions regarding the optimum path that data should take from its source to its ultimate destination across the network. Routers make these decisions based on logical addressing information found at layer 3 of the OSI model. Routers can perform much more complex functions than bridges or layer 2 switches, but doing so results in more overhead and slower performance. It should be noted, however, that routers are faster at layer 3 functions than the equivalent layer 3 switches. For more information on layer 2 and layer 3 switching, refer to Cisco Systems' Cisco IOS 12.0 Switching Services (Cisco Press, ISBN 1578701570).
Multiplexor (MUX). A MUX is a device that takes a broadband network segment with high bandwidth and divides it into multiple segments with less bandwidth. An inverse multiplexor (I-MUX) does the opposite: It joins multiple low-bandwidth segments to form a single high-bandwidth broadband link.
The wiring and the connecting devices form the basic network infrastructure. Connected to this basic infrastructure are the many individual computer systems that must use the network. These systems are called hosts because each one has the potential to host data that is of interest to the other computer systems on the network. Although it is a connecting device in applications, a router is also a host on a network. In fact, any network device with a physical or logical address on the network is considered a host. This includes all internetworking devices above the level of an unintelligent hub at the Physical OSI layer. Hosts perform many different functions on the network:
In order for hosts to make use of the network infrastructure for communications, they must first agree to a set of rules or protocols. They might need to adopt a number of different protocols to allow multiple types of hosts to communicate with each other. Much more information on protocols can be obtained in the Network+ books referenced earlier in this chapter.
There are protocols that specify how to use the physical media to carry binary data. There are protocols that specify how to get that data from one system to another. There are also protocols that specify how to present the data to the particular host that receives it.
In the existing infrastructure, you will find that a number of network services are currently being offered:
Network management services
IP address services
Host name resolution services
These network services make use of the network infrastructure and define the functionality of the network itself.
When designing a new network infrastructure, you need to determine the following:
Which network services are currently in use?
Which of the currently used network services will be retired?
How will implementation of the new design impact the functionality of the currently used network services?
What new network services will be added to the network in the future?
As you did with the applications currently in use, you should list all the network services that are currently in use by the organization. Include in your list the specific network requirements for each service. For example, the DNS name resolution service requires the TCP/IP protocol. After you list all the network services currently in use and the requirements for each service, consult with company management to determine how critical each service is to the business. Whenever you face conflicts with the requirements of network services, this priority level can help you resolve them or determine where to make compromises.
The TCP/IP suite is the standard network protocol used by nearly every network in existence today. It is the default protocol used by Microsoft Windows 2000, as well as Windows NT 4.0. Many other network operating systems use TCP/IP. Myriad network devices support the TCP/IP protocol suite. A network that is based on the TCP/IP protocol has certain elements that must be considered carefully at the design stage in order for the network to operate effectively and efficiently:
The IP addressing scheme
The IP address assignment process
The host name registration process
The host name resolution process
These TCP/IP network elements must be carefully considered when creating your new network infrastructure design. We will discuss each of them in the following sections.
The IP Addressing Scheme
When you're setting up an IP network, one of the fundamental building blocks is the IP addressing scheme. Each host connected to the network must have a unique address in order for it to communicate with other hosts on the network. You have two options when creating an addressing scheme. The first is to use a public IP address assigned by a public entity. This is the option you must take if you are connecting your company to the Internet and do not have any kind of proxy server or firewall between your company's network and the Internet. The second option is to use the so-called reserved private IP addresses. The private IP addressing space is a set of IP addresses that are reserved and cannot be used in the public network. These IP addresses give you the ability to configure as many subnets and hosts as you require, regardless of the public IP assigned to your company.
The private or reserved IP addresses adhere to the following ranges. The first private network range is from 10.0.0.0 through 10.255.255.254. The second private network range is from 172.16.0.0 through 172.31.255.254. The third private network range is from 192.168.0.0 through 192.168.255.254.
Whichever addressing scheme you choose, public or private, one main concern is that your addressing strategy be a planned and organized one rather than haphazard. Standards should be developed for assigning IP addresses to devices. Set aside a range of numbers that specify certain devices. For instance, printers might fall within the host range of 25 to 46; routers might fall within the address range for hosts of 15 through 25; switches might be in the host address range 5 through 15. There are no hard and fast rules here. Just be sure that the approach you take is organized and meaningful. Always make sure that your addressing scheme is documented.
The IP Address Assignment Process
After you have arrived at an IP addressing scheme that is organized and works well throughout the enterprise, you will need to decide how to get these IP addresses assigned to each individual station on the network. This could be a manual process, which would require a lot of administration and hands-on technical work. If you take a manual approach to IP address assignment, a technician must approach each device on the network and configure its IP address. This can be very time-consuming and expensive and also introduces the possibility of human error. This will, in turn, lead to troubleshooting costs and downtime.
Another approach, which might save time and money, is to automate the assignment of IP addresses to end stations. One way of doing this is to use the Dynamic Host Configuration Protocol (DHCP). When using DHCP, you first configure a DHCP server with an address pool containing all the IP addresses that you can use to assign to hosts on your network. When DHCP clients first power up and connect to the network, they request an IP address from the server. They do this by broadcasting a DHCP request over the network. All the DHCP servers on the network seeing this request will respond with an "offer" of an IP address. The client will accept the first offer it receives, and it will lease that address for a fixed period of time.
The time frame during which a workstation may use the address it received from DHCP server is called the lease duration. The lease duration is a configurable parameter. Configure a long lease duration if your clients tend to stay in one place and the network topology doesn't change often. However, if your network topology changes frequently, or if you do not have an address pool with enough addresses to sufficiently meet the demand for them, configure a shorter lease duration.
The Host Name Registration Process
TCP/IP end stations are not identified by only their IP address. For the sake of human convenience, they are also recognized by a host name.
NetBIOS Computer Names Versus DNS Host Names
Although they are created for the same reason, NetBIOS computer names in downlevel clients are not the same as DNS host names. Most downlevel clients have the ability to have two different "friendly" names associated with their IP address. One is the NetBIOS or computer name, and the other is the DNS host name.
For more information on NetBIOS over TCP/IP, NetBIOS naming, and WINS name resolution, refer to New Riders' Networking with Microsoft TCP/IP, Third Edition, by Drew Heywood (ISBN 0735700141).
Most people find that names are easier to remember than IP addresses. In addition, although IP addresses might be dynamically assigned and changed frequently for the same host, the host name remains consistent. So, regardless of whether the host has a new IP address or changes its IP lease frequently, you can consistently refer to the host by name, and connectivity will be ensured. It is tempting to use interesting (even wild) host names for your end stations. Individual end users might want to use a host name that reflects their personality or favorite hobbies. However, this is probably not the best approach to assigning IP host names across your enterprise, because it makes it very difficult for a user to determine what resources reside on which host. Develop a standard naming convention that is meaningful, hierarchical, and organized, and that you can apply throughout the enterprise. Once you have determined your host name scheme, you will need to deploy those names across the enterprise. This typically requires a manual visit to each end station for configuration.
The Host Name Resolution Process
Once host names have been assigned to each end station on your TCP/IP network, you will need some way to translate host names into the IP addresses associated with those end stations. The service that provides this function in a TCP/IP network is the domain name service (DNS).
Using DNS, you first configure a database on a DNS server. The database will contain all the host names for each of the end stations in your enterprise, as well as the IP addresses that are associated with each end station. When a workstation wants to establish connectivity with another, it will first request a lookup from the DNS server. It will specify the name of the host to which it is trying to connect. The DNS server will respond by providing the IP address associated with that name. Once the connecting station receives the IP address, it will begin to communicate directly with the host associated with that IP address.
In the past, the creation of the DNS database was a manual process. This required a DNS administrator to spend many laborious hours determining the IP addresses assigned to each of the hosts in the enterprise. This was further complicated by the use of DHCP protocol, because the IP addresses associated with each of the host names in the database change from time to time, perhaps frequently. Frequent changes to the IP address assignments require frequent changes to the DNS database. This can be very time-consuming and expensive, and it also introduces the opportunity for human error.
A solution to this problem has recently been developed and is being implemented in enterprises today. This solution is called dynamic DNS (DDNS). When using DDNS, clients that receive their IP address assignments through DHCP protocol communicate their host name back to the DHCP server. The DHCP server, in turn, communicates with the DNS server, indicating that an address assignment has been made as well as the host name associated with that address assignment. Through this process, the DNS database is updated dynamically each time an address assignment is made. This approach minimizes the administrative intervention required in assigning IP addresses and host names. It allows the DNS database to be kept up to date without manual intervention and minimizes the troubleshooting associated with IP address changes when the DNS database is not kept up to date.
Once you have configured your DNS servers and the database is up to date, each individual host will need to be configured to communicate with the DNS server to resolve host names. If you are using the DHCP protocol for a dynamic host IP address assignment, you can also configure an additional parameter that will specify the DNS server that the host should use for IP host name resolution.
It is important to note that no matter what you include in your network infrastructure design, it will be completely useless if the hardware in place cannot support it. You will need to take an inventory of the hardware in the existing network infrastructure. Examine the inventory information for the hardware configuration of each of the components.
Upgrades and Rollouts
Earlier in this chapter, we discussed the need to list and prioritize the applications currently in use by the company. You should also have examined each of the applications to determine its network requirements. Where you might have found conflicting requirements between applications, the criticality of each application helped you arrive at a compromise. Two additional issues regarding applications that must be considered are application upgrades and rollouts.
If an upgrade to an existing application is available, the company might want to consider implementing the upgrade at the same time it implements the new network infrastructure. You might find that company management is inclined to combine as many changes to the environment as possible when implementing a new infrastructure. This is certainly not the wisest approach to take, because it complicates the troubleshooting process when issues arise during the implementation. Nevertheless, it is a common approach. You will need to become aware of any company plans to upgrade existing applications.
Upgrades that occur at the same time as a new network infrastructure implementation are not always bad. In fact, in the case of legacy applications or situations where there are conflicts in application requirements, an upgrade might be just the resolution you are looking for. Upgrading legacy applications might allow you to discontinue the use of older, less-efficient protocols. When two or more applications have conflicting requirements, upgrading one or more of them to a newer version might resolve the conflict. Of course, it is always possible that an upgrade to one application might introduce new conflicts in network requirements that must be resolved before your design can be implemented.
Technical Support Structure
An issue that needs to be considered when analyzing the impact of a network infrastructure design is the company's ability to support it after it is implemented. A major component of the total cost of ownership for the network infrastructure is the ongoing cost to support that infrastructure. It is important to take the time to examine the organization's technical support structure to determine if it can effectively support the new network infrastructure. Many infrastructure plans overlook support. Although support is often a post- implementation concern, preparing for it should be given a higher priority in the planning phase of the project.
Here are some questions you should ask:
Do company employees perform technical support, or is it outsourced to contractors?
Does the technical support structure effectively address the support needs of the existing network infrastructure?
Are there new technologies in the new network infrastructure design that would pose new challenges to the existing technical support structure?
How should the technical support structure be modified or enhanced in order to most effectively support the new network infrastructure?
You might find that the existing technical support structure is overworked and struggling to meet the support demands of the existing network infrastructure. This is frequently the case. Meet with company management, as well as the technical support staff, and try to determine the reason for this situation. Usually such circumstances are related to budgetary issues. Try to tactfully point out the shortcomings of the existing technical support structure to company management. While being mindful of office politics and general customer service principles, you will want to convey the shortcomings of the group without demeaning the members themselves. Point out to management how these shortcomings can be resolved by the addition of personnel or by an investment in the training of the existing staff.
Network and Systems Management
A major component in the supportability of a network infrastructure is the investment in network and systems management tools for monitoring the health of the network infrastructure components, as well as the individual systems that reside on the network. These are essential for minimizing downtime and troubleshooting costs.
Numerous tools are available for performing network and systems management. Depending on the devices and systems that comprise the company's network, you might find one or more of these tools currently in use.
When examining the company's existing network and systems management tools, you need to determine the following:
Do the existing tools for network and systems management effectively provide support for the existing network infrastructure?
Is the organization currently making the most effective use of the available tools for network and systems management?
How will the implementation of the newly designed network infrastructure impact the effectiveness of the network and systems management tools currently in use?
What changes to the systems and network management tools currently in use will need to be made in order to support the new network infrastructure?
Which additional network and systems management tools will need to be implemented in order to effectively support the new network infrastructure?
This part of the chapter analyzed the physical makeup of a network and the impact that various systems and services have on the infrastructure design. Applications are often the most important piece of a company's infrastructure, because they are directly used by the end users for production purposes and the company's line of business. These applications can be prioritized by a number of factors, including criticality, resources required, time sensitivity, and visibility/user perception.
Network infrastructure includes all workstations, servers, interconnection devices, and even physical wiring. This book can't go into the depth that this material could cover. Instead, it gives you an overview of physical devices and cabling that can be researched further in some of the recommended texts, such as the CompTIA Network+ books referenced earlier in this chapter. These devices include cabling, such as thinnet, thicknet, UTP, and fiber optic, as well as network devices, such as repeaters, hubs, bridges, switches, routers, gateways, and multiplexors. Another good overview reference for networking devices and protocols is Chapter 7 of A+ Fast Track (New Riders Publishing, ISBN 0735700281).
Network services were also touched on in this part of the chapter. These network services make use of the network infrastructure and define the functionality of the network itself. As an integral part of the network infrastructure, proposed changes to these will also need to be documented and analyzed for impact to the network infrastructure. These services include directory services, file and print services, Web services, management services, name resolution services, dynamic IP addressing, and many others.
IP addressing also was covered briefly in this section. This overview discussed IP addressing, address assignment, name registration, and name resolution. Particular attention should be paid to address assignment and name registration throughout this book. This portion of the TCP/IP infrastructure is the most affected by updates included in Windows 2000, especially DDNS. References were given for supporting books on TCP/IP due to this book's focus and space limitations. Entire libraries of books have been written on this most popular Internet protocol.
Hardware upgrades, systems support, and network systems management will all have a profound impact on the network infrastructure. After analysis, a good design for these support areas and tactics will have an extremely positive impact on the network infrastructure design.
This topic of support leads us directly into the next section on analyzing client computer access requirements.