LAN Troubleshooting (Networking)

LANs often experience performance problems that can be attributed to faulty hardware. Indications of faulty hardware are that network performance is poor or that one or more devices cannot communicate. One way to identify hardware problems on a network is to track communication errors. Errors occur on a network when a device receives a transmission but is unable to make sense of it.

Types of Errors

On an Ethernet network, for example, several distinct types of errors can appear, including:

COLLISIONS These occur when two devices attempt to place a packet on the network at the same time. Collisions are detected when the signal on the cable is equal to or exceeds the signal produced by two or more transceivers that are transmitting simultaneously.

JABBERS This is a frame that is greater than 1518 bytes and has a bad CRC (checksum value). If a transceiver does not halt transmission after 1518 bytes, it is considered to be a jabbering transceiver.

FRAGMENTS These are packets that appear at a receiving station, contain fewer than the required minimum of 64 bytes, and have a bad CRC. Fragments are generally caused by collisions.

OVERSIZE These are frames that are larger than the maximum 1518 bytes and have a good CRC.

UNDERSIZE These are frames that contain less than the minimum of 64 bytes and have a good CRC.

CRC/ALIGNMENT ERRORS These are packets that do not contain the proper CRC. In addition, if the frame does not end on an 8-bit boundary, an alignment error will occur. Both of these types of errors are grouped and counted as CRC/alignment errors.

Tools

If a technician could identify what types of errors occur on the network, along with where they occur, the problems could be resolved more easily. This requires the use of special tools, such as RMON probes, that collect performance data on the network segments, as well as a monitoring application to interpret the results.

Briefly, a typical troubleshooting procedure begins by opening the monitoring application’s statistics window for a RMON probe. For the selected segment, the items of interest for an Ethernet network are CRC/alignment errors, undersizes, oversizes, fragments, jabbers, and collisions. A network segment that has faulty hardware would show significant errors of one or more of these types.

Once it is determined that a network segment has excessive errors of a particular type, it might be helpful to track the error to a single device or a few devices. A technician can accomplish this by opening the Top N window of the monitoring application, which reveals the top 10 devices transmitting errors (Figure 70). The device at the top of the graph generates the most errors and thus requires the most attention.

At this point, a problem with errors in the network traffic has been discovered and the type of error or errors has been determined, as well as the possible source for the errors. The next step is to find out what is causing the errors.

If collisions are high, it might mean that the network segment is being overused and that some workstations need to be reassigned to other less-used segments. Another cause of high collision rates are cable segments that are too long.

If there is jabber, a faulty transceiver is the cause. The transceiver is a chip, which is responsible for transmitting bits to and receiving bits from the information transfer, the station reinserts the token on the ring. A token-holding timer controls the maximum amount of time a station can occupy the network before passing the token to the next station.

Star

A star network has a central node that connects to each station by a single, point-to-point link (Figure 68). All communications between nodes the medium. Transceivers are typically located on the network interface card (NIC). Eliminating jabber entails replacing the transceiver, which usually means replacing the entire NIC.

The star topology.

Figure 68

The star topology.

Figure 70

Hewlett Packard’s HP NetMetrix/Win, a RMON-based network monitoring tool, displays the top 10 devices transmitting errors. The device at the top of the graph generates the most errors and thus requires the most attention when troubleshooting network performance problems.

Hewlett Packard's HP NetMetrix/Win, a RMON-based network monitoring tool, displays the top 10 devices transmitting errors. The device at the top of the graph generates the most errors and thus requires the most attention when troubleshooting network performance problems.

Packet fragments are generally the result of collisions, and the two types of errors usually go hand in hand. Taking corrective actions for collisions eliminates fragments as well.

Illegal length frames—oversizes and undersizes—can easily be traced to the transmitting station, since the frame is well formed and contains the source address in the header. Such frames are generally caused by a faulty LAN driver. The driver may be out of date or the file on disk may be corrupt.

CRC/alignment errors are usually due to cabling problems. Any one of the following cabling problems can cause this type of error:

■ Segment too long

■ Damaged cable

■ Segment not properly grounded

■ Improper termination

■ Taps too close

■ Noisy cable (electromagnetic interference)

■ Faulty controller cards

Last Word

Network troubleshooting is usually more involved than what is discussed in this section, and it varies with token ring, FDDI, and other types of networks. Regardless of the type of network, the speed and efficiency of troubleshooting is dependent on proactive measures, such as maintaining baseline measurements of network parameters, e.g., usage and errors. With performance parameters for comparison, network administrators can more easily detect when something has gone wrong, ascertain the probable cause, and take corrective action.

Next post:

Previous post: