Home > Articles > Operating Systems, Server > Microsoft Servers

  • Print
  • + Share This

Major Network Components

This section provides a brief discussion of the basic components of most networks—repeaters, bridges, hubs, and routers.


In the most basic sense, networks work by sending electrical signals across cables. These signals, however, attenuate as they traverse each cable; at some point along a cable, the signal may fail to carry any further. Devices called repeaters are used in digital networks to regenerate signals as they go down cables so that they can reach destination hosts. Repeaters are often used to increase the physical size of LANs, allowing additional systems and peripherals to connect to a preexisting LAN. Note, however, that Ethernet does not need repeaters, because of the limited distances over which traffic to segments travels.

Contrasted with other types of network components, repeaters are in many respects more passive in their functionality. In the most elementary sense, they simply take signals, magnify them, and then send them along a network cable without the capacity to selectively filter, in any way, what is sent. A potential problem in connection with repeaters, therefore, is that a repeater can potentially overwhelm a network with volumes of traffic. Hosts in any part of a network in which repeaters are present can produce an enormous amount of traffic volume even though some, much, or all of this traffic may be superfluous to the systems along the way.

You might think, therefore, that repeaters would be frequent targets of network attacks in the Windows NT or any other networking environment. Gaining unauthorized access to a repeater or sending network traffic to a repeater in a manner that causes the repeater to fail or other possible attack scenarios could conceivably lead to widespread denial-of-service. Attacks against other components of networks (for example, routers and firewalls) are, however, generally not only easier to remotely perpetrate, but they are also more likely to subvert higher-level network functionality, such as routing.


A bridge is very similar in functionality to a repeater, but operates at Layer 2 of the OSI model. Therefore, a bridge actually filters traffic transmitted over the network based on the Layer 2, or MAC, address. Bridges dynamically update their routing tables with source addresses as they receive packet traffic. They determine the MAC address for each machine on the basis of the contents of packets (the MAC addresses of both the source and destination hosts) sent over the network. The destination MAC address of each packet is then used to selectively filter traffic packets in the following manner:

  • Packets with unidentified destinations are sent on to every network segment to which the bridge is connected.

  • Packets with identified destinations on other network segments to which the bridge is connected are sent on to the segment on which the destination machine resides.

  • Packets in which the source and destination address are both within the same network segment are not sent to any other network segment.

Bridges are both good and bad in terms of their contribution to the problem of managing networks. An advantage of bridges (at least compared to repeaters) is that the former are not as passive; they can at least send traffic to other network segments or, if appropriate, keep traffic within a particular local segment. They can also selectively filter traffic on the basis of each packet's destination MAC address. A disadvantage is that bridges cannot filter broadcasts (transmissions from hosts that are intended to reach other computers independently of the computers' addresses).


Some types of networks have network cables that go from the connected computers to one single point. Hubs (sometimes called concentrators) connect the cables at this point. Hubs vary in sophistication. Some of them just rebroadcast any signals they receive from a particular cable on to every other cable. Others are higher end, in that they function as network switches. They determine packets' destinations and then resend the signal exclusively to the appropriate cable over which the packets must be transmitted to reach this destination.


Routers are the final network component to be considered in this chapter. Routers, like bridges, route and screen traffic; but whereas bridges operate at Layer 2 and therefore use the MAC address to determine whether to forward traffic, routers operate at Layer 3, and therefore forward traffic based on the network layer address. In addition, routers have knowledge of networks outside the networks to which they are directly connected, and consequently can interconnect one LAN to many others. By contrast, bridges have no intelligence other than to determine which network a certain MAC address exists on. They can be used only to interconnect the networks to which they are directly connected. Whereas bridges can interconnect only networks that use the same data link layer protocol, routers can connect networks with different data link layer protocols, as long as they use the same network layer protocol. The important result is that routers can handle packets on a higher, more abstract level rather than information supplied by any NIC. Consider, for example, the effect of dynamic IP addressing. One machine is assigned one particular IP address for a while, and then a different address sometime afterward. Routers have no particular difficulty dealing with dynamic addressing provided the assigned IP addresses are legitimate and routing tables have been set up and maintained suitably.

The basis for routing, therefore, is routing tables. When a router boots, it has few addresses in its table. Routing protocols, such as the Address Resolution Protocol (ARP), add entries to the tables as packets pass through. Network administrators can also add or delete entries in routine tables as desired. Routers may also support Access Control Lists (ACLs), rule lists specifying whether to allow or deny inbound (and also often outbound) traffic according to certain variables. These variables include source and/or destination IP address, type of protocol (for example, Simple Mail Transfer Protocol [SMTP] and File Transfer Protocol (FTP), and type of packet (SYN, ACK, FIN, and so forth). A router's ACL may block all incoming packets destined for a particular IP address, for example. A bridge, on the other hand, sends and filters packets on the basis of the network segment to which a machine with a certain MAC address is connected.

Domain Name Server

When you send a mail message to an Internet address or hit a Web site, a mechanism needs to translate this address (for example, http://security.globalintegrity.com) into a numeric IP address. The service that performs this task (called address resolution) is domain name service (DNS). Using its database of addresses and IP addresses, the Address Resolution Protocol (ARP) is used to resolve layer three (IP) addresses to layer 2 (MAC) addresses. Similarly, the Reverse Address Resolution Protocol (RARP) is used to get an IP address from a MAC address. Windows NT supports ARP, but not RARP. DNS does not use ARP at all. DNS resolves a Fully Qualified Domain Name (FQDN) to an IP address, using a forward lookup query. To get an FQDN from an IP address, a reverse lookup is performed. Because so many Internet applications and services depend on DNS, internetworking as we know it would be virtually impossible without this service.

The Internet consists of a large number of computing domains; each Internet-connected host is part of one domain. The host security.globalintegrity.com, for example, is part of a domain called globalintegrity.com. A given DNS server can resolve addresses only within its own domain, but it also knows how to forward resolution requests to other DNS servers if it cannot resolve the name. If unable to resolve globalintegrity.com, the DNS servers within a tree forward the request all the way to the top (the root domain) to find the one that knows about the type of organization (for example, .com, .org, .gov, .edu). From there the request is forwarded to a DNS server that knows .com addresses, and then to still another that knows globalintegrity.

DNS serves as a convenient target for perpetrators for numerous reasons. At a minimum, its mechanisms are automatic and are not selective concerning the

particular host allowed to make queries. One of the most common attacks on DNS servers is "cache poisoning," in which the perpetrator replaces part or all of the name cache it builds with bogus entries.

Many vendors' routers also support additional functionality, such as encrypting network traffic sent from one router to another. From a security standpoint, routers are potentially very useful network components. On the other hand, if an intruder gains unauthorized access to or causes disruption of a router, the consequences can be extremely unpalatable for security. An intruder could, for example, change a router's ACLs or set the routing configuration to cause traffic to be sent over a fixed route over the network. Ensuring that this fixed route is used is important in a number of different types of network attacks, including attacks in which an attacker has installed a network traffic capture device on one of the machines along this route.

Because the Internet switches packets instead of using circuits to move packets from one point to another, we often refer to it as a packet-switching network. At this point, you might assume networking consists of an orderly flow of packets from one point through one more routers and finally to the destination point. However, this is not true. Packets may, for example, be too large for a router to handle. In this case, IP routers typically simply discard the packets. In other cases they are broken into fragments. When these fragments reach their destination, they are usually reassembled into a continuous data stream. The process of fragmentation and reassembly is normally transparent to users. In TCP/IP, a transmission's packets do not go over fixed routes, but instead look for the best, most efficient route available. Different packets from the same transmission might therefore go over somewhat different routes from the sending to the receiving host. Packets may arrive out of sequence; the receiving host normally can reconstruct the original order.

  • + Share This
  • 🔖 Save To Your Account

Related Resources

There are currently no related titles. Please check back later.