The Switched Environment

Frame Forwarding

Switching as a General Concept in Networking and Telecommunications

The concept of switching and forwarding frames is universal in networking and telecommunications. Various types of switches are used in LANs, WANs, and the public switched telephone network (PSTN). The fundamental concept of switching refers to a device making a decision based on two criteria:

•     Ingress port
•     Destination address

The decision on how a switch forwards traffic is made in relation to the flow of that traffic. The term ingress is used to describe where a frame enters the device on a port. The term egress is used to describe frames leaving the device from a particular port.

When a switch makes a decision, it is based on the ingress port and the destination address of the message.

A LAN switch maintains a table that it uses to determine how to forward traffic through the switch. Click the Play button in the figure to see an animation of the switching process. In this example:

• If a message enters switch port 1 and has a destination address of EA, then the switch forwards the traffic out port 4.
• If a message enters switch port 5 and has a destination address of EE, then the switch forwards the traffic out port 1.
• If a message enters switch port 3 and has a destination address of AB, then the switch forwards the traffic out port 6.
  

The only intelligence of the LAN switch is its ability to use its table to forward traffic based on the ingress port and the destination address of a message. With a LAN switch, there is only one master switching table that describes a strict association between addresses and ports; therefore, a message with a given destination address always exits the same egress port, regardless of the ingress port it enters.

Cisco LAN switches forward Ethernet frames based on the destination MAC address of the frames.

Dynamically Populating a Switch MAC Address Table

Switches use MAC addresses to direct network communications through the switch to the appropriate port toward the destination. A switch is made up of integrated circuits and the accompanying software that controls the data paths through the switch. For a switch to know which port to use to transmit a frame, it must first learn which devices exist on each port. As the switch learns the relationship of ports to devices, it builds a table called a MAC address, or content addressable memory (CAM) table. CAM is a special type of memory used in high-speed searching applications.

LAN switches determine how to handle incoming data frames by maintaining the MAC address table. A switch builds its MAC address table by recording the MAC address of each device connected to each of its ports. The switch uses the information in the MAC address table to send frames destined for a specific device out the port which has been assigned to that device.

A switch populates the MAC address table based on source MAC addresses. When a switch receives an incoming frame with a destination MAC address that is not found in the MAC address table, the switch forwards the frame out of all ports (flooding) except for the ingress port of the frame. When the destination device responds, the switch adds the source MAC address of the frame and the port where the frame was received to the MAC address table. In networks with multiple interconnected switches, the MAC address table contains multiple MAC addresses for a single port connected to the other switches.

The following steps describe the process of building the MAC address table:
1.The switch receives a frame from PC 1 on Port 1 (Figure 1).

Figgure 1.

2. The switch examines the source MAC address and compares it to MAC address table.

·         If the address is not in the MAC address table, it associates the source MAC address of PC 1 with the ingress port (Port 1) in the MAC address table (Figure 2).

·         If the MAC address table already has an entry for that source address, it resets the aging timer. An entry for a MAC address is typically kept for five minutes.

 Figure 2.

3. After the switch has recorded the source address information, the switch examines the destination MAC address.

·         If the destination address is not in the MAC table or if it’s a broadcast MAC address, as indicated by all Fs, the switch floods the frame to all ports, except the ingress port (Figure 3).

Figure 3.

4. The destination device (PC 3) replies to the frame with a unicast frame addressed to PC 1 (Figure 4).

Figure 4.

5. The switch enters the source MAC address of PC 3 and the port number of the ingress port into the address table. The destination address of the frame and its associated egress port is found in the MAC address table (Figure 5).

Figure 5.

6. The switch can now forward frames between these source and destination devices without flooding, because it has entries in the address table that identify the associated ports (Figure 6).

Figure 6.

Switch Forwarding Methods

As networks grew and enterprises began to experience slower network performance, Ethernet bridges (an early version of a switch) were added to networks to limit the size of the collision domains. In the 1990s, advancements in integrated circuit technologies allowed for LAN switches to replace Ethernet bridges. These LAN switches were able to move the Layer 2 forwarding decisions from software to application-specific-integrated circuits (ASICs). ASICs reduce the packet-handling time within the device, and allow the device to handle an increased number of ports without degrading performance. This method of forwarding data frames at Layer 2 was referred to as store-and-forward switching. This term distinguished it from cut-through switching.

Store-and-Forward Switching

Store-and-forward switching has two primary characteristics that distinguish it from cut-through: error checking and automatic buffering.

Error Checking

A switch using store-and-forward switching performs an error check on an incoming frame. After receiving the entire frame on the ingress port, as shown in the figure, the switch compares the frame-check-sequence (FCS) value in the last field of the datagram against its own FCS calculations. The FCS is an error checking process that helps to ensure that the frame is free of physical and data-link errors. If the frame is error-free, the switch forwards the frame. Otherwise the frame is dropped.

Automatic Buffering

The ingress port buffering process used by store-and-forward switches provides the flexibility to support any mix of Ethernet speeds. For example, handling an incoming frame traveling into a 100 Mb/s Ethernet port that must be sent out a 1 Gb/s interface would require using the store-and-forward method. With any mismatch in speeds between the ingress and egress ports, the switch stores the entire frame in a buffer, computes the FCS check, forwards it to the egress port buffer and then sends it.

Store-and-forward switching is Cisco’s primary LAN switching method.

A store-and-forward switch drops frames that do not pass the FCS check, therefore does not forward invalid frames. By contrast, a cut-through switch may forward invalid frames because no FCS check is performed.

Cut-Through Switching

An advantage to cut-through switching is the ability of the switch to start forwarding a frame earlier than store-and-forward switching. There are two primary characteristics of cut-through switching: rapid frame forwarding and fragment free.

Rapid Frame Forwarding

As indicated in the figure, a switch using the cut-through method can make a forwarding decision as soon as it has looked up the destination MAC address of the frame in its MAC address table. The switch does not have to wait for the rest of the frame to enter the ingress port before making its forwarding decision.

With today’s MAC controllers and ASICs, a switch using the cut-through method can quickly decide whether it needs to examine a larger portion of a frame’s headers for additional filtering purposes. For example, the switch can analyze past the first 14 bytes (the source MAC address, destination MAC, and the EtherType fields), and examine an additional 40 bytes in order to perform more sophisticated functions relative to IPv4 Layers 3 and 4.

The cut-through switching method does not drop most invalid frames. Frames with errors are forwarded to other segments of the network. If there is a high error rate (invalid frames) in the network, cut-through switching can have a negative impact on bandwidth; thus, clogging up bandwidth with damaged and invalid frames.

Fragment Free

Fragment free switching is a modified form of cut-through switching in which the switch waits for the collision window (64 bytes) to pass before forwarding the frame. This means each frame will be checked into the data field to make sure no fragmentation has occurred. Fragment free mode provides better error checking than cut-through, with practically no increase in latency.

The lower latency speed of cut-through switching makes it more appropriate for extremely demanding, high-performance computing (HPC) applications that require process-to-process latencies of 10 microseconds or less.