Data Link Layer Devices
Objective:
- Describe the purpose and functions of various network devices
At the Data Link layer, either a bridge or a Layer 2 switch can be installed to segment a LAN. Hubs and repeaters at the Physical layer only serve to extend a network. With segmentation, switches and bridges create a separate collision domain for each connected node, which effectively reduces the number of collisions that occur on that network.
Remember from Chapter 1 that a collision domain is a group of nodes that shares the same media and are segmented by switches or bridges. A collision occurs if two nodes attempt a simultaneous transmission within the same collision domain. This reinforces the need for an increased number of collision domains. Figure 3.7 demonstrates how a bridge creates two collision domains.
Figure 3.8 provides an example of a situation in which a switch creates separate collision domains.
EXAM ALERT
Know what a collision domain is and that a bridge and/or switch will segment a network and create an additional collision domain for each segment. Routers not only segment collision domains, but they also segment broadcast domains.
Bridges
Bridges were created to alleviate several expansion-related network issues. As networks were growing and becoming more complex, hubs and repeaters no longer provided sufficient network resources. Because they do not segment the network, all the devices connected to a hub or repeater had to share the same bandwidth. Also, if one device sent a frame it could collide with a frame from another device on that LAN. This meant that all devices on that LAN had to take turns sending frames. Again, this is not very efficient as additional devices are added to a network.
Transparent bridges were introduced and helped solve these growing pains. The word transparent is used to indicate that the other devices on a network are not aware of its existence. Bridges use a software application to forward frames.
The following are the primary tasks performed by both bridges and switches:
- The source MAC address of every inbound frame is examined to learn its MAC address.
- Frames may either be forwarded or filtered depending on the destination MAC address (they can also be flooded if the destination is unknown).
- Eliminates loops that are caused by redundant connections by configuring Spanning Tree Protocol (STP).
Learned MAC addresses and their interfaces are stored in a bridge table on the bridge or switch. When a new frame arrives on that bridge or switch, the device refers to the bridge table to decide how to forward or filter the frame. If the frame’s destination MAC address is on a different segment of that LAN, the device forwards the frame to that segment. If the frame’s destination MAC address is on the same segment as the source MAC address, the device filters the frame. That frame reaches its destination without the assistance of a bridge or switch. Figure 3.9 shows a segmented LAN with the MAC addresses of each end user.
As frames are received by the bridge or switch from each end user, it updates its bridge table with their MAC addresses and the interface on which the frame came into the device. Table 3.8 shows the bridge table of this bridge.
If the incoming frame destination address is…
- Unicast—The bridge checks the bridge table first. If the destination unicast address is not in the bridge table, it forwards the frame to all interfaces except for the interface that originally sent the frame. If the destination unicast address is in the bridge table and on a different interface than the interface that originally sent the frame, it forwards the frame. If the destination unicast address is in the bridge table and on the same interface as the sender, the frame is filtered.
- Multicast—The bridge forwards the frame to all interfaces except for the interface that originally sent the frame.
- Broadcast—The bridge forwards the frame to all interfaces except for the interface that originally sent the frame.
Using the same diagram and your new bridge table, I will give you a source and destination address. Please fill out whether the frame will be filtered or forwarded. If it is forwarded, also fill out the outbound interface to which the frame will be sent.
EXAM ALERT
Bridges and switches examine the source MAC address of each inbound frame to learn MAC addresses.
Switches
Layer 2 switches are multi-port bridges; therefore, they have all the same functionality of bridges. There are differences that differentiate a switch from a bridge. For example, switches utilize hardware or Application-Specific Integrated Circuit (ASIC) chips to forward frames rather than software. Also, each port of the switch has a dedicated bandwidth. If the dedicated port on a switch is 10Mbps, the connected LAN segment has a dedicated bandwidth of 10Mbps. This works in the same manner for 100Mbps and 1000Mbps dedicated switch ports. This feature also sets a switch apart from a bridge that has a low port density.
EXAM ALERT
For the test, know that switches are multi-port bridges that use ASIC hardware chips for frame forward ing. Dedicated bandwidth enables the switch port to guarantee the speed assigned to that port. For example, 100Mbps port connections get 100Mbps transmission rates.
A popular ethernet switch port is the 10/100 ethernet port, where you can set the port to pass traffic at 10Mbps or 100Mbps. Chapter 6, “Introduction to Cisco Routers and Switches,” goes into more detail regarding specific Cisco devices, including the 2950 series switches.
Duplex
It is important that you understand duplex logic and how it affects traffic on a network. The communication mode of a device may either be half-duplex or full-duplex, depending on the connection type.
Half-duplex allows for one-way communication, which means that a device can only send or receive a data transmission at any given time. This option does not allow for simultaneously sending and receiving data. As part of a shared collision domain, hubs are inherently set up for half-duplex. Bandwidth suffers because a collision detection technology such as the CSMA/CD algorithm must be implemented. Collision detection can chew up 50–60% of the bandwidth on that ethernet LAN.
Full-duplex allows for two-way communication, which means that a device can simultaneously send and receive data transmissions. Full-duplex is available with dedicated switch port connections to a single device. If a switch port connection is configured for full-duplex, the CSMA/CD algorithm must be disabled. An ethernet connection set for full-duplex allows for 100% transmission speeds in both directions. For example, a 100Mbps connection can transmit data simultaneously at 100Mbps in each direction.
With ethernet, if a switch port and NIC offer multiple speed options as well as half- and fullduplex settings, autonegotiation can be configured on both devices. The switch and NIC automatically negotiate the connection speed and duplex so that the settings on both ends match. You may have heard of a 10/half or 100/full connection before. The term 10/half refers to a 10Mbps half-duplex connection. It is more likely that you will see 100/full, which indicates a 100Mbps full-duplex connection.
Autonegotiation may not always be a reliable option. There have been some instances where the switch port goes into error disable mode because of massive errors. Configuring or “hard coding” the port and NIC to the appropriate speed and duplex settings may resolve the issue when the port is reactivated.
Hubs use half-duplex technology. Switches can be set up for full-duplex.
Microsegmentation
Microsegmentation occurs when a switch creates a dedicated path for sending and receiving transmissions with each connected host. Each host then has a separate collision domain and a dedicated bandwidth.