1. Symptoms
At a chemical trading company in East China, after upgrading the network from 10Mbps to 100Mbps, approximately half of the workstations on the same floor failed to achieve increased speed. Meanwhile, five workstations on another floor couldn’t connect to the network. Some workstations on both floors experienced slower speeds than before, though not universally, and only certain servers or workstations showed decent access speeds. The company lacked network maintenance tools, so they relied on indirect monitoring of network traffic and collision rates through software, observing elevated collision rates of up to 20% in some micro-segments but were unsure how to address the issue.
According to Lucy, who is responsible for network management, all workstations could access the network before the upgrade, with only some stations having speed issues that were manageable. The company’s network was relatively small, spanning two and a half floors, with 280 workstations. Three workgroup switches were positioned in the computer room, providing connections to the respective floors. These switches were interconnected via a 100Mbps hub. A single router connected to a frame relay network through the workgroup switches. Below the switches, a star topology connected the link interfaces to user desktops.
The upgrade process was straightforward, involving the replacement of 10Mbps switches with 100Mbps switches, and 10Mbps hubs with 100Mbps hubs, with the equipment layout on the racks remaining largely unchanged. Workstations were equipped with 100Mbps network cards, and the upgrade was conducted during non-business hours on a weekend. Although spot checks on some workstations indicated satisfactory performance after the upgrade, issues emerged when regular business operations resumed on Monday.
2. Diagnostic Process
The network structure appeared simple, and the issues manifested as three common symptoms: some stations couldn’t access the network, some experienced slower speeds, and half the stations couldn’t achieve the expected 100Mbps speed after the upgrade. These are typical problems encountered during network upgrades, often referred to as “network upgrade symptoms.”
Initially, a network testing tool (F683 network tester) was connected to the micro-segment of the non-connecting stations to observe network behavior. The tool couldn’t detect these workstations, but intermittent responses were observed during ping tests. These symptoms often suggest mismatches between the workstations and the network, which can be caused by protocol mismatches, driver issues, network card speed mismatches, link pulse polarity mismatches, physical parameter mismatches, and the use of incorrect cable types (such as Cat-3 cables). Testing interfaces directly using network testing tools, network fault finders, or network analyzers can help identify and address these issues.
Further tests were conducted on the five workstations with connectivity issues, revealing that the network adapters were set to auto-negotiate speeds, while the switch ports were manually set to 100Mbps, creating a mismatch. Replacing the connectors with Cat-5 connectors, testing with Cat-5 cabling, and using 100Mbps network adapters resolved the connectivity problems.
Tests on workstations experiencing slower speeds after the upgrade showed that the collision rate was high, and the network operated at a speed of only 10Mbps, even though the switch settings were set to 10/100Mbps auto-negotiation. To investigate further, cable links were examined and found to be consistently substandard. Replacing the connectors with Cat-5 connectors and testing with Cat-5 cabling helped address the issues.
Additionally, a daisy-chain effect was discovered on another floor with four groups of 17 workstations, leading to significant collision issues and FCS frame errors. The network was suffering from delays and traffic collisions, causing a reduction in network speed. Some users experienced improved speeds, but it didn’t reach the expected 100Mbps.
3. Diagnostic Recommendations
It is recommended to conduct comprehensive cable system testing, adjust switch configurations, and eliminate daisy-chain topologies to enhance network performance.
4. Afterword
After two weeks, all problematic connectors were replaced, and testing confirmed the cable system was in good condition. This situation occurred because different contractors worked on different floors, using substandard connectors for some of the connections. A check of the daisy-chain topology found seven groups with “suspected” configurations. Following the recommended solutions, four additional workgroup switches were added to alleviate the daisy-chain effect. The network is now operating effectively.