Skip to content

Ethernet switching

A switch routes packets to different devices connected to its ports. It usually does switching at layer 2, which is on the same level as MAC addresses.

IP addresses are layer 3

Layer 2 switching operations

Address learning

  • Learn source MAC address of each data frame transmitted and keeping track of the port where the frame enters the switch

Flooding

  • Send out on all ports when the destination MAC address is unknown or a broadcast

If destination address is unknown to the switch, it forwards the frame to all segments except the one which sent the frame

Forwarding

  • Learns source MAC address of each data frame that is transmitted
  • Notes that port where the frame enters the switch
  • Forwards when the destination MAC matches an entry in the MAC Address Table

If the destination address is on a different segment, the switch forwards the frame to the appropriate segment

Filtering

  • Filters when destination is located on the same port

If the destination address is on the same segment as the frame, the switch blocks the frame from going on to other segments

Switching modes

Cut through switching

  • Fastest - used by default
  • Reads beginning of frame up till destination MAC address (6 bytes)
  • Bad frames are forwarded, may congest the network
    • Fragment < 64 bytes
    • Corrupted frame - FCS check error

Store and forward switching

  • Reads entire frame into buffer and checks for errors before forwarding
  • Reduces bandwidth wastage as bad frames are not forwarded

Fragment-Free switching

  • Fragment free switching holds the frame until the first 64 bytes are read from the source to detect a collision before forwarding
  • Reduces latency as the frame is sent after it has read 64 bytes

Switching Modes Summary

Modes Advantages Disadvantages
Store-and-Forward No invalid frames forwarded Variable latency (delay in switch)
Cut-through Low constant latency Forward fragments & error frames
Fragment-Free Filters out fragments Forward error frames

Redundancy in switch topologies

  • Resilience with multiple switches and trunk links
  • One link or device fails another takes over

Problem with Redundancy

  • Broadcast storms
    • Floods broadcast through non-source ports and so on
  • Multiple frame transmission
    • Multiple copies of frames may be delivered to the destination
  • Inconsistent switch tables
    • Switch tables per switch is inconsistent causing it to send the frame through the wrong interface

Redundancy without loops

  • One path at a time
  • Redundant path is close but ready to be open when needed
  • Process is quick and automatic
  • Spanning Tree Protocol does this

Spanning Tree Protocol (STP)

4 steps:

  1. Select root bridge
    1. Choose one with the lowest bridge ID -> Priority + MAC address
  2. Select root port
  3. Select ports on other switches with the lowest distance to root bridge
  4. Select designated ports
  5. Block unused ports

Spanning Tree Overview

  • STP is a Layer 2 link-management protocol that provide path redundancy while preventing undesirable loops
  • Loop-free topology is accomplished when a switch recognizes a loop and blocks one or more redundant ports automatically
  • Bridge Protocol Data Units (BPDUs) are used by switches in a network to exchange information regarding their status
  • STA to resolve and shutdown redundant paths
  • STP is defined in the IEEE 802.1d specification

Spanning Tree Protocol Summary

  • Redundant Paths is good for network resilience
  • Broadcast Storms caused by redundant paths
  • STP overcomes broadcast storms by using Loop Avoidance:
    • Redundant paths are blocked
    • One active path between any node in the network
    • If active path fail, blocked path is activated automatically
    • Network resilience is achieved without broadcast storms

Root bridge election

  • Switches exchange BDPUs with each other, containing their bridge ID.
    • Bridge ID = Priority (default: 32768 = \(2^15\)) + MAC address
  • Switch with the lowest bridge ID is elected as new bridge

Routing table

For directly connected networks

Destination network Next hop / outgoing interface
192.168.2.0 Directly connected

Stub networks

  • Send all non local traffic out a single interface

Cisco IOS commands

Show spanning tree:

show spanning-tree

Show IP addresses & on/off status of interface:

show ip interface brief

Show all configuration of the router:

show running-config

show run all

Configure interface IP:

config terminal
interface <e>
ip address <ip> <subnet mask>

Configure static routing: 

ip route <network> <subnet mask> <outgoing interface>

Outgoing interface can only use P2P, using next hop IP can be used for multipoint

Configure stub routing: 

ip route 0.0.0.0 0.0.0.0 <outgoing interface>

Last update: June 11, 2023
Created: June 11, 2023