Using Network testing equipment to troubleshoot network failures

Network management often encounter some difficult problems, such as slow or poor performance. This time, if there is a network tester to help detect failures, will undoubtedly greatly improve the efficiency.

So how to use Fluke Company's Lanmeter network tester to detect slow network or poor performance?

Run the network Statistics (network Stats) test to check for high network utilization and abnormal high collision rates. Apply a small amount of background traffic (100 frames per second LLC traffic, 100 bytes per frame) and run the test again. If you find that collisions or FCS errors increase as the transmission increases, you should run a cable scan (Cable Scan).

If the number of collisions is very high (more than 5%), run the Crash Analysis (collision) test to determine how much bandwidth is lost due to collisions. If the loss of bandwidth caused by local and remote collisions is added, the average greater than 0.5% is required for fault diagnosis. At the same time, check whether the collision is "sudden", and the increase of sudden collision does not increase with the increase of network traffic. That is, if there is a significant change in the number of collisions without a corresponding increase in traffic, there may be a serious physical layer problem somewhere. Flow level and collision must have some kind of relationship. If the number of collisions is high at an acceptable level of network performance, there may be too many sites in the collision domain to transmit, or the network structure needs to be optimized to reduce the distance between network sites. Excessive collisions are often caused by physical media such as no or incorrect termination, impedance discontinuity (bad connectors, cable connections, extruded cables, etc.), or bad network adapters.

If the utilization rate is high (sustained spikes over 60%) and the number of collisions is acceptable (the average is less than 5%), then the network may be saturated. This is not possible because Ethernet segments typically have a higher collision rate if they are highly utilized. When the utilization rate is close to 100%, the number of collisions will be much higher than the number of good frames. You may need to install a switch, bridge, or router to separate the network into small enough subnets to support the traffic load.

Run Network statistics (network Stats) test and check for errors (other than collisions). If an error occurs, run the error statistics (Stats) test and Zoom in (Zoom) to understand the MAC address of the problematic site, and then use an expert test (EXPERT-T) to isolate the specific problem or check the cable with a cable scan (Cable). If the problem is intermittent, try replacing the NIC that is suspected to be problematic because these problems may not be displayed during testing.

Check to see if the server or service that the user is connecting to is at the far end of the WAN or connected by routing. If so, use the ping test (NetWare ping,icmp ping) to check the response through the network segment and compare the results to your benchmark or hearing test results. To ensure that the frame is not lost, try to ping continuously, and check the number of requests and the number of responses is equal (the IP host will often fail on the first ping, and then need to try more times). If the quantity is inconsistent, the remote medium or interconnect device (bridge, router, etc.) may be full load or problematic. Dropping frames is also a symptom of a cable failure, such as a UTP cable, or a geodetic loop along the same axis. If you suspect a drop frame, run a cable scan (Cable Cable), test the slow site (and the corresponding file server) to connect to the network cable, or connect to any bridge or router and the cables between them.