A Brief Discussion of Forwarding Methods in Network switch

2025-11-15 16:12:29

1. Store-and-Forward Switching (SNS): Network switch operating in store-and-forward mode read the entire frame of data into memory and check its correctness before sending information. Although this method takes longer than the pass-through method, it stores and forwards data, ensuring accuracy. Because Network switch operating in store-and-forward mode do not propagate erroneous data, it is more suitable for large local area networks.

2. Error Control: After receiving a complete frame at the ingress port, the Industrial grade switch compares the Frame Check Sequence (FCS) of the last field of the datagram with its own FCS. The FCS check process helps ensure that the frame has no physical or data link errors. If the frame check is correct, the Core switch forwards it; otherwise, the data is discarded.

3. Automatic Buffering: Store-and-forward Network switch support mixed connections of Ethernet with different speeds by buffering at the ingress port. For example, receiving a frame sent at a 1Gb/s rate and forwarding it to a 100Mbps Ethernet port requires store-and-forwarding. When the ingress and outgress port speeds do not match, the Core switch puts the entire frame content into the buffer, calculates the FCS check, forwards the output buffer, and then sends the frame.

4. Cut-through Switching

One advantage of cut-through switching is its speed compared to store-and-forward technology. Network switch using cut-through mode read the frame header and decide which port to send the data to before receiving the entire data packet, without buffering data or checking data integrity. This switching method has two main characteristics: fast frame forwarding and invalid frame handling.

5. Fast Frame Forwarding

Once the Core switch finds the destination MAC address in the MAC address table, it immediately makes a forwarding decision without waiting for the rest of the frame to enter the port.

Network switch using cut-through mode can quickly determine whether it is necessary to check more parts of the frame header for additional filtering purposes. For example, the Core switch can check the first 14 bytes (source MAC address, destination MAC address, Ethernet type field) and the following 40 bytes to implement IPv4 Layer 3 and Layer 4 related functions.

6. Invalid Frame Handling

For most invalid frames, Core switch using cut-through mode do not discard them; erroneous frames are forwarded to other network segments. If a high error rate (invalid frames) occurs in the network, cut-through switching may adversely affect bandwidth, and corrupted and invalid frames can cause bandwidth congestion. In congested situations, these Network switch will buffer the frames like store-and-forward vehicles.

7. Fragmentless Forwarding

Fragmentless forwarding is an improvement on the cut-through method. Before forwarding, each Core switch checks if the frame is larger than 64 bytes (if smaller, it is discarded) to ensure there are no fragmented frames. Fragmentless forwarding offers better error detection than cut-through without actually increasing latency. It is well-suited for high-performance computing applications, i.e., applications where process-to-process latency is less than 10 milliseconds.