How to Efficiently Maintain a Local Area Network
When there is a fault, check the network card first.
During the Local Area Network (LAN) maintenance process, it's common to encounter network disconnection issues. When such problems arise, we should first carefully check if the settings of the network cards in all machines connected to the network are normal. To do this, you can sequentially open the "Control Panel/System/Device Manager/Network Adapter" settings window using your mouse. In this window, check for any conflicts in interrupt numbers and I/O addresses (it's best to set the interrupts on each machine to be the same for easier comparison). Continue until the property of the network adapter displays "This device is working properly," and you can find at least your own machine in "My Network Places," indicating that the network card configuration has no issues. LAN management software can manage the LAN effectively, making LAN management more convenient and efficient.
Confirm that the network cables and network devices are working normally.
After checking and finding no issues with the network card, we can try to view other computers on the network through "My Network Places." If you still cannot see other machines on the network, this may indicate a problem due to interrupted network connections. Network connection faults usually include internal breaks in the network cable, poor contact with twisted-pair cables or RJ-45 connectors, or quality issues with the network connection equipment itself, or improper connections. At this point, we can use a cable tester to check if the lines are broken, then use a substitution method to test the quality of the network equipment. If there are no issues with the network cables and network cards themselves, we should check if there are any reasons related to software settings, such as incorrect interrupt numbers potentially causing malfunctions.
Check if the driver program is intact.
After checking the hardware and confirming it's correct, check if the driver program itself is damaged. If not damaged, verify if it is installed correctly. If these can be judged as normal and there are no device conflicts but still cannot connect to the network, at this time, you can delete the network adapter in the system configuration, then restart the computer. The system will detect new hardware and automatically search for and install the driver program. The author has used this method multiple times during the installation of Windows 98 peer-to-peer networks to solve the issue of "unable to access the internet."
Correctly configure the network card.
If after confirming there are no issues with the network medium, you still cannot connect, return to the network card settings. Check for any device resource conflicts. Many times, conflicts do not always prompt warnings. Possible device resource conflicts include: NE2000-compatible network cards conflicting with COM2 both using IRQ3 (Realtek RT8029), PCI Ethernet network cards and graphics cards both "liking" IRQ10. To resolve such device conflicts, we can follow these steps: First, in the settings window, disable COM2 and forcibly set the network card interrupt to 3; if encountering a conflict between PCI interface network cards and graphics cards, we can solve it by not assigning an IRQ to the graphics card, which means setting the "Assign IRQ for VGA" item in the CMOS to "disable."
Disable the PnP function of the network card.
Some network cards support PnP functionality, but after installation, they may not work well or even fail to function. For this reason, we can resolve this fault by disabling the PnP function of the network card. To disable the PnP function of the network card, you must run the network card's setup program (usually found in the driver package). After starting the setup program, enter the setup menu. Disable the PnP function of the network card and modify the IRQ setting to a fixed value. Save this setting and exit the setup program. If no other devices occupy this IRQ, you can ensure that there will be no IRQ conflicts. Additionally, if installing Windows 95/98 operating systems, you must ensure that the operating system does not treat the corresponding interrupt type as having PnP functionality. Therefore, in the "CMOS," change the interrupt type from "PCI/ISA PnP" to "Legacy ISA." Using this method can resolve most PnP network card device conflicts, but it may not be effective for all PCI network cards because some network cards' setup programs do not provide the option to disable PnP functionality.
Reasonably configure the server's hard drive.
Users who use a LAN for office work often use the network to print materials and access files, requiring LAN maintenance. Due to certain reasons, the speed of network access might be abnormal, leading us to mistakenly believe that the cause of reduced network speed could be bottlenecks in some devices within the network, such as network cards, switches, hubs, etc. However, the biggest factor affecting network speed is still the speed of the server's hard drive. Therefore, correctly configuring the hard drive of the server in the LAN will greatly improve the overall network performance of the LAN. Typically, when setting up the hard drive, we need to consider several factors:
The hard drives in the server should ideally be fast-spinning and have large capacity, because faster spinning hard drives allow for quicker data access over the network;
The hard drive interface in the server should preferably be SCSI, as this interface transmits data faster than IDE or EIDE interfaces, adopting parallel data transmission mode for sending and receiving data;
If conditions permit, we can install a RAID card on the network server, as RAID cards can significantly enhance the read/write performance and security of hard drives;
It's also important to note that on the same SCSI channel, low-speed SCSI devices (such as CD-ROMs) should not be shared with hard drives, otherwise the high-speed data transmission capability of the hard drive's SCSI interface will not be fully utilized.
Correctly use "bridge" devices
"Bridge" devices are typically used for network devices within the same subnet, whereas routers are used for different segments of the network. At my workplace, we once installed a microwave networking device. After the physical devices were connected, we went online to debug, but the server kept prompting that the current subnet number should be the subnet number of the other party. After changing the subnet number of the server to match the other side, the server's warning disappeared. Ah! It turned out this was a set of bridge-type devices. Later, when installing microwave networking equipment at another location, we switched to products from another manufacturer. Before connecting, we had already changed the subnet numbers on both sides to be consistent, but after installing the equipment, the server again showed an alarm: incorrect current route. After modifying the subnet number on one side, the alarm disappeared. From this, it can be seen that correctly distinguishing between "routing" devices and "bridge" devices is crucial when setting network parameters.
Connect according to rules
Each computer in the LAN is connected using twisted-pair cables, but simply connecting two computers with twisted-pair cables doesn't necessarily achieve communication purposes. We must connect according to specific rules. I once tried to connect two computers more than 100 meters apart with twisted-pair cables to enable communication, but no matter how hard I tried, I couldn't connect successfully. Later, after consulting an expert, I learned that the connection distance of twisted-pair cables cannot exceed 100 meters. Additionally, if we need to connect two computers more than 100 meters apart, we must use conversion devices, and when connecting conversion devices and switches, we must perform crossover wiring. This is because in Ethernet, generally two pairs of twisted-pair cables are used, arranged in positions 1, 2, 3, and 6. If the original paired wires are separated and used individually, it will result in crosstalk, thus producing significant interference (NEXT), which greatly affects network performance. In a 10M network environment, this situation isn't obvious, but in a 100M network environment, if the traffic is heavy or the distance is long, the network will be unable to connect.
Strictly enforce grounding requirements
During LAN maintenance, weak signals are transmitted. If operations are slightly inappropriate or not conducted according to the specific operational requirements of network devices, interference information may appear in the network, severely causing the entire network to disconnect; especially for some network transition devices, since they involve remote lines, their grounding requirements are very strict. Otherwise, the network device will not reach the specified connection rate, resulting in various inexplicable faults during the networking process. Once, I unintentionally plugged the power cord of the router into a municipal electricity outlet, and the 128k DDN dedicated line was unable to connect to the Internet. The telecommunications bureau checked the lines and found them normal, but finally, upon checking the zero-ground voltage of the router's power supply, it was found to be incorrect. After switching back to the UPS outlet, everything returned to normal. Another time, the grounding end of the router's power plug was damaged, causing frequent data packet loss. During ping connections, it was intermittently good and bad. After replacing the power cord, everything returned to normal. Therefore, when using network devices, we must operate under the specified conditions of the device; otherwise, it will cause great trouble for our work.
Use high-quality, fast new network cards
During LAN maintenance, it's common for computers to be unable to communicate. There could be many reasons for this. I once counted that most faults in the LAN were related to network cards, either because the network card was not correctly installed, or the network cable had poor contact, or the network card was old and could not be correctly recognized by the computer, or because the network card installed on the server could not withstand the impact of large amounts of data and eventually broke down. Therefore, to avoid such phenomena, we must invest adequately. If the network card is installed on a server, we must use high-quality network cards because servers generally run uninterruptedly, and only high-quality network cards can "work" for long periods. Additionally, since servers transmit large amounts of data, the capacity of the network cards we purchase must match it, achieving the principle of "a good horse deserves a good saddle."
Reasonably configure the switch
Switches are important data exchange devices in LANs, and correctly and reasonably using switches can significantly improve data transmission performance in the network. I once configured a switch port to 100M full-duplex and installed an Intel 100M EISA network card on the server. Everything was normal after installation, but during high-traffic data transmission, the speed became extremely slow. Finally, it was discovered that this network card did not support full-duplex. After changing the switch port to half-duplex, the fault disappeared. This shows that the port of the switch and the speed and duplex mode of the network card must be consistent. Currently, there are many self-adaptive network cards and switches. In theory, they should correctly adapt to speed and duplex modes, but in practice, due to differences in brands, full-duplex mode often cannot be correctly achieved. Even though the server network card is set to full-duplex, the duplex light on the switch does not turn on, and only manual forced settings can resolve the issue. Therefore, when setting network device parameters, we must refer to the network device parameters on the server or other workstations, trying to make each device work in a matched manner.
