Perhaps you've spent the past several weeks preparing for your next certification exam, studying complex exam-question situations and the troubleshooting steps that help you choose the correct multiple-choice answer. This approach might serve you well in the exam room, but in the real world, sometimes you need to examine the basics.
During the past few years, I've encountered several unusual troubleshooting situations in which cabling was the ultimate culprit. You might have heard of situations in which network cabling ran up an elevator shaft, where it fell prey to the electrical interference of the elevator's motor. Perhaps you've heard the story about the server that failed once a week, always on the same night. As it turns out, a cleaning person was unplugging the machine to plug in a vacuum.
IT professionals are quick to spread such stories across the Internet when others are to blame. These same professionals are probably less eager to pass on any tales that paint the troubleshooter in a less favorable light.
Perhaps the easiest problem to address is a broken tab at the end of a patch cable. If you have a crimp tool and can match the wires in the correct color order, you can quickly replace a patch cable. Other fixes, such as taping the cable in place, are far from ideal and invite failure—invariably at the worst possible time. Network cables that are too long can also pose problems on your network. Long and insufficiently twisted cables can result in cross-talk interference with network transmissions.
Another common problem can occur when you depend too heavily on network devices to auto-sense connection speeds and duplexes. The connection might not "shake hands" at full speed. In the worst-case scenario, a link that depends on auto-detection can jump between speeds, never connecting properly. If you encounter such problems, be sure to check your hardware's documentation.
Other common cabling troubles result from using crossover cables instead of straight-through cables, and vice versa. Interestingly, some of the latest switches can auto-detect cable configurations, which lets you use either cable type. But to ensure that you avoid problems, you should confirm that you have the right connectors at both ends and use a quality cable tester to check for shorts and breaks in the line.
Cable trouble—particularly when SCSI cables are involved—can also result in strange server behavior. Because the SCSI bus provides a linear connection between the SCSI card and SCSI devices, the cable should be terminated at both ends. This advice seems simple enough, but many devices can act as terminators themselves. Connecting terminated devices in the middle of the cable makes communications to other devices unreliable at best.
Another common SCSI termination problem might occur if you connect a 50-pin device to a newer 68-pin controller. If the device you're connecting is the only device on the SCSI chain, you should have no problem. If, however, the device shares the chain with 68-pin devices, problems might arise. First, the slower device slows performance on the entire chain. Second, the 68-pin-to-50-pin adapter must terminate the 18 unused pins; some adapters can't do this, and as a result, performance suffers for the 68-pin device.
Finally, don't use traditional 40-pin IDE cables to connect your drives to current-generation IDE controllers, even though they do fit. Use the new 80-pin cables, which place a ground wire on every other wire. These new cables keep the signal clean and provide faster, error-free communications. Until next time, stay connected, and keep on running!