1. Symptoms
Today is my first round of check-ups, and the patients are complaining that most of their internet connections at the sites are slower than before the system upgrade. Some sites experience intermittent connectivity issues, while others can’t connect at all. Initially, they were using 10 Mbps Ethernet, which was very stable and had excellent performance. After upgrading to a 100 Mbps system, these symptoms appeared. The number of users hasn’t increased, and no large or multimedia applications have been enabled.
I’ve tried reinstalling system and application software, reconfiguring the servers and sites, virus scanning, and resetting all networking devices, but none of these actions have worked. Interestingly, two machines work normally when moved to another location, and when laptops are connected to the problematic hub ports, they work fine. Even after changing the two network cables, the issue remains. Restoring the machines to their original location and replacing the network cards (originally 3COM cards) resolved the problem, though the cause remains unclear. Since most Ethernet users are unable to work, we’ve temporarily reverted to the 10 Mbps Ethernet system.
2. Diagnosis Process
Observing from the network management of the 10 Mbps system, the average network traffic is 3%, which is significantly lower than 40%. Considering that no large or multimedia software is running, there shouldn’t be any significant “discount” in terms of speed. Connecting FLUKE’s F683 network tester to the hub port, the network traffic is measured at 35%. The collision rate is at 23%, far exceeding the healthy standard of 5%. The reported error types include delay collisions, FCS frame errors, and a few local errors. It can be reasonably concluded that there is a severe issue with the cabling system.
Testing the cables reveals that, except for the two cables at the testing points, all other cabling links have attenuation and near-end crosstalk that do not meet the standards. Testing these cables with a Category 3 standard shows that they all pass. Examining the cable packaging, the presence of “Lucent Cat5” suggests counterfeit products.
The lengths of the links for the two workstations are 78 meters and 86 meters, and testing their network card ports reveals insufficient signal amplitude, and there is no built-in 3COM manufacturer label on the instrument.
3. Conclusion
Many counterfeit products are labeled as Category 5 when they are actually covered with Category 3 cables. Users won’t experience application issues in a 10 Mbps Ethernet environment, but as soon as they upgrade to a 100 Mbps environment, only a few short links can barely function. For the two sites that worked normally after being moved, it was discovered that the link length was only 3 meters, and it was standard Category 5 cable (typically used for temporary testing at this site). In the original location, the test lengths were 45 meters and 37 meters. Due to the weak signal transmission capability of the network card, there was significant signal attenuation in the 100 Mbps system, causing difficulties in internet access. When connected with a 3-meter cable, the impact of attenuation was minimal, allowing for normal internet access. Network card testing revealed that they were counterfeit cards.
4. Postscript
ne month later, the ‘patients’ called to inform me that after rewiring, the network was working perfectly, with no noticeable interruptions, even when running some multimedia software. The unfortunate news is that the original system integrator is nowhere to be found, so we couldn’t make a claim.