Basic of switching Method of switching Functions of bridging

Bridges and switches are layer 2 devices that segment (break up) collision domains. A collision domain basically includes all the devices that share a media type at layer 1.

Difference between bridge and switch

Functions	Bridges	Switches
Form of switching	Software	Hardware
Method of switching	Store and forward	Store and forward, cut-through, Fragment-free
port	2-20	100 plus
Duplex	Half	Half and full
Collision domains	1 per port	1 per port
Broadcast domains	1	per vlan
STP instances	1	1

Methods of Switching

Store and Forward

Store and Forward is the basic mode that bridges and switches use. It is the only mode that bridges can use, but many switches can use one or more of the other modes as well, depending on the model. In Store-and-Forward switching, the entire frame is buffered (copied into memory) and the Cyclic Redundancy Check (CRC), also known as the FCS or Frame Check Sequence is run to ensure that the frame is valid and not corrupted.

Cut Through

Cut Through is the fastest switching mode. The switch analyzes the first six bytes after the preamble of the frame to make its forwarding decision. Those six bytes are the destination MAC address, which, if you think about it, is the minimum amount of information a switch has to look at to switch efficiently. After the forwarding decision has been made, the switch can begin to send the frame out the appropriate port(s), even if the rest of the frame is still arriving at the inbound port. The chief advantage of Cut-Through switching is speed; no time is spent running the CRC, and the frame is forwarded as fast as possible

Fragment-free

Switching will switch a frame after the switch sees at least 64 bytes, which prevents the switching of runt frames. This is the default switching method for the 1900 series. 2950 doesn’t support cut-through Fragment-Free switching is sometimes called "runtless" switching for this reason. Because the switch only ever buffers 64 bytes of each frame, Fragment Free is a faster mode than Store and Forward, but there still exists a risk of forwarding bad frames, so the previously described mechanisms to change to Store and Forward if excessive bad CRCs are received are often implemented as well.

Functions of Bridging and Switching

Learning

Address learning refers to the intelligent capability of switches to dynamically learn the source MAC addresses of devices that are connected to its various ports. These addresses are stored in RAM in a table that lists the address and the port on which a frame was last received from that address. This enables a switch to selectively forward the frame out the appropriate port(s), based on the destination MAC address of the frame. Anytime a device that is connected to a switch sends a frame through the switch, the switch records the source MAC address of the frame in a table and associates that address with the port the frame arrived on.

Bridges place learned source MAC addresses and their corresponding ports in a CAM (content addressable memory

Forwarding

Address learning refers to the intelligent capability of switches to dynamically learn the source MAC addresses of devices that are connected to its various ports. These addresses are stored in RAM in a table that lists the address and the port on which a frame was last received from that address. This enables a switch to selectively forward the frame out the appropriate port(s), based on the destination MAC address of the frame. Anytime a device that is connected to a switch sends a frame through the switch, the switch records the source MAC address of the frame in a table and associates that address with the port the frame arrived on.

There are some situations in which a switch cannot make its forwarding decision and flood the frame.
Three frame types that are always flooded:

• Broadcast address Destination MAC address of FFFF.FFFF.FFFFF
• Multicast address Destination MAC addresses between 0100.5E00.0000 and 0100.5E7F.FFFF
• Unknown unicast destination MAC addresses The MAC address is not found in the CAM table

Removing layer-2 loops

Spanning Tree Protocol (STP - 802.1d) The main function of the Spanning Tree Protocol (STP) is to remove layer-2 loops from your topology

Static MAC Addresses

In addition to having the switches learn MAC addresses dynamically, you can manually create static entries. You might want to do this for security reasons. Statically configuring MAC addresses on the switch is not very common today. If configured, static entries are typically used for network devices, such as servers and routers.

Port Security Feature

Port security is a switch feature that allows you to lock down switch ports based on the MAC address or addresses associated with the interface, preventing unauthorized access to a LAN. Three options are possible if a security violation occurs—the MAC address is seen connected to a different port.

• protect
  When the number of secure addresses reaches the maximum number allowed, any additionally learned addresses will be dropped.
• restrict
  Causes the switch to generate a security violation alert.
• shutdown
  Causes the switch to generate an alert and to disable the interface. The only way to re-enable the interface is to use the no shutdown command. This is the default violation mode if you don’t specify the mode.

EtherChannels

An EtherChannelis a layer 2 solution that allows you to aggregate multiple layer 2 Ethernet-based connections between directly connected devices. Basically, an EtherChannel bundles together multiple Ethernet ports between devices, providing what appears to be single logical interface.

EtherChannels provide these advantages:

• Redundancy If one connection in the channel fails, you can use other connections in the channel.
• More bandwidth each connection can be used simultaneously to send frames.
• Simplified management Configuration is done on the logical interface, not on each individual connection in the channel.

EtherChannel Restrictions

Interfaces in an EtherChannel must be configured identically: speed, duplexing, and VLAN settings (in the same VLAN if they are access ports or the same trunk properties) must be the same.

When setting up EtherChannels, you can use up to eight interfaces bundled together:

• Up to eight Fast Ethernet connections, providing up to 800 Mbps
• Up to eight Gigabit Ethernet connections, providing up to 8 Gbps
• Up to eight 10-Gigabit Ethernet connections, providing up to 80 Gbps

You can have a total of six EtherChannels on a switch.

EtherChannel Operations

Channels can be formed dynamically between devices by using one of two protocols: Port Aggregation Protocol (PAgP) or Link Aggregation Control Protocol (LACP). Remember that ports participating in a channel must be configured identically. Once a channel is formed, load balancing can be used by the connected devices to utilize all the ports in the channel. Load balancing is performed by reducing part of the binary addressing in the frame or packet to a numeric value and then associating the numeric value to one of the ports in the channel.

Load balancing can use MAC or IP addresses, source or destination addresses, or both source and destination address pairs. With this fashion, you are guaranteed that all links in the channel will be utilized; however, you are not guaranteed that all the ports will be utilized the same. For example, if you are load balancing based on source addresses; you are guaranteed that different source MAC addresses will use different ports in the channel. All traffic from a single-source MAC address, however, will always use the same port in the channel. Given this situation, if you have one device generating a lot of traffic, that link will possibly be utilized more than other links in the channel. In this situation, you might want to load balance based on destination or both source and destination addresses.