[ Pobierz całość w formacie PDF ]
contain a hexadecimal number assigned by the manufacturer thatÇs unique to each piece of equipment.
Just as you need to know someoneÇs telephone number to call them, computers must know each otherÇs
addresses to communicate. Depending on the protocol in use, various addressing schemes are used. For the
exam, you should be aware of TCP/IP and IPX addressing schemes (see Chapter 12 for more information).
In a workstation, the MAC address is usually burned into the NIC card. On a router, each port has its own
MAC physical address. Theoretically, no two devices ever have the same MAC address. Although, we have
heard of instances where this has occurred in a network with very unpleasant circumstances resulting.
44
IÇm known as IP, but my friends call me MAC
Hardware (MAC) addresses are used to get data from one local device to another. However, not all network
operating systems (NOS) use the physical address to reference network nodes. This sets up the conflict
between the network (logical) address and the MAC (physical) addresses.
Network operating systems assign a logical network name to each networked device, such as
ACCTG_SERVER, NT1, or FRED, to make it easy for its human users to reference its resources. On the
other hand, references on the network itself, that is those on Layer 1 (Physical layer), use the physical
addresses provided by the Data Link layer to reference the actual devices on the network. When you request
services from the file server FRED, a service like DNS (Domain Naming System) or WINS (Windows
Internet Name Service) is used to translate or resolve the node name FRED into its logical address, which is
typically an IP address. In some cases, a HOSTS or LMHOSTS file may be used instead to resolve the node
name to its logical address. The Data Link layer activities then resolve the logical address into its
corresponding MAC address. To resolve between these two addresses involves a process called (what else?)
address resolution, which associates logical network addresses to physical MAC addresses, and vice versa.
Remember The protocol for this service is ARP (Address Resolution Protocol). ARP maintains a small
database in memory, called the ARP cache, that cross-references physical and logical addresses. When a
device wants to communicate with a local device, it checks its ARP cache to determine whether it has that
deviceÇs MAC address. If it doesnÇt, it sends out an ARP broadcast request, as shown in Figure 4-1, to all
devices on the local network. Each device examines the message to see whether the request is intended for it.
If it is, the device responds with its MAC address, which is stored in the sending deviceÇs ARP cache. In the
example shown in Figure 4-1, USER1 wants to communicate with FRED, a file server. However, USER1
doesnÇt have a MAC address in its ARP cache for FRED, so it sends out a broadcast message that asks FRED
to respond with its MAC address, which FRED does.
When a workstation or server needs to communicate with a device remote to the local network, essentially
the same process takes place with the exception that a router through which the remote device is accessed will
likely respond with its MAC address and not that of the device itself.
Controlling access to the network
The primary media access control mechanism defined in 802.3 for use in the Data Link layer is CSMA/CD
(Carrier Sense Multiple Access/Collision Detection) access method. CSMA/CD is the method used in
Ethernet networks for controlling access to the physical media by network nodes. As its name infers,
CSMA/CD (say it ten times fast to lock it away in your brain!) tries to keep network devices from interfering
with each otherÇs communications by detecting access attempts by multiple devices. When sneaky devices
avoid detection, and they do, CSMA/CD detects and deals with the collision that undoubtedly occurs.
45
Figure 4-1: To resolve an unknown address on the network, a broadcast request is sent out.
Avoiding collisions
To avoid collisions, CSMA/CD devices ÝlistenÛ or sense signals on the network backbone before sending a
message over the network. If the network is quiet, meaning itÇs not in use, the device can send a message.
Otherwise, the device waits until the network isnÇt in use. However, if between the time a device decides the
network is available and the time it actually transmits its message, another device sends a message, the two
messages may collide on the network. When this happens, the device that detected the collision sends out an
alert to all network devices that a collision has occurred. All devices quit transmitting for a random amount of
time to clear the line.
Remember The CSMA/CD process can be described as follows:
1. Listen to see whether the wire is being used.
2.
If the wire is busy, wait.
3.
If the wire is quiet, send.
4.
If a collision occurs while sending, stop, wait a specific amount of time, and send again.
When a collision is detected by a sending device, it sends out a jamming signal that lasts long enough for all
nodes to recognize it and stop broadcasting. Then each device waits a random amount of time to begin the
CSMA/CD process again. This amount of time is determined by a back-off algorithm that calculates the
amount of time the device should wait before resuming its attempts to transmit.
Working on a busy intersection
A collision domain is a network segment in which all devices share the same bandwidth. The more devices
[ Pobierz całość w formacie PDF ]