On any network, there are several causes of congestion that prevent data from flowing freely. Lisa Bock covers troubleshooting to discover the cause of a slow network that includes low bandwidth, outdated hardware, and unmanaged multicast traffic.
- [Instructor] Networks need to be available nearly 100% of the time. A single failure should not significantly impact network performance. The network should be able to grow to meet the ever-changing demand for more hosts on the network. No matter how good the design of the network is, the network administrator must be able to manage the complexity of the network. When things go wrong, we begin a troubleshooting process.
When troubleshooting, there are several approaches. The first is generally gathering information and ask detailed questions. For example, if someone calls to report the network is down, ask is it just you, the entire office, or the entire wing? Once you've asked those questions, you're able to then narrow the scope and define the problem. You'll want to consider possible causes and then develop a solution along with an alternative.
Then you'll need to implement the solution and then test. If you don't get the results that were expected, you'll need to revert back to the original state and try other solutions. Once you're confident the network is up and operational, you'll document the solution so as to prevent further problems. On any network, there are several causes of congestion that prevents data from moving freely. Low bandwidth is where the amount of data that can flow is reduced as multiple hosts contend for the same bandwidth.
Although there are many reasons, the main culprits include voice over IP and videoconferencing, cloud backup services, and Facebook, Pandora, and YouTube. On a network, a broadcast is one to everyone on the network, and those broadcasts can include the protocols ARP and DHCP. Too many hosts in a broadcast domain can overload a network.
Outdated hardware that doesn't process quickly enough or have ample buffer space can slow traffic and cause congestion. Multicast is one to many that are listening. A multicast header doesn't have a destination IP address. It's a multicast address, which is a Class D IP Version 4 address. An unmanaged switch will send the packet out to every network port, and that multicast will be treated as a broadcast.
If multicast traffic is all over the network, you should check the configuration on your devices to see if IGMP is properly monitoring multicast traffic. Tune the devices so that multicast traffic is forwarded only to the devices that should be listening. On any network, there are several causes of congestion that prevent data from flowing freely. Some of the causes include low bandwidth, outdated hardware, and unmanaged multicast traffic.
- Tapping into the network
- Baselining the network
- Troubleshooting to discover the cause of a slow network
- Merging traffic
- Sanitizing packet captures
- Capture engines
- Optimizing packet captures
- Basic and advanced IO graphs
- TCP stream graphs