Networking Basics CCNA 1 Chapter 8

1
Networking Basics CCNA 1Chapter 8
2
Ethernet Switch Operations

• Layer 2 Bridging and Switching Operations
• Earliest networking devices were repeaters and
  hubs
• Multiple LAN segments could be connected to make
  larger LANs, within 5-4-3 design rules
• As it became apparent that reducing size of
  collision domains was important, bridges were
  created
• Bridges are aware of Ethernet framing and Layer 2
  MAC addressing (IEEE 802.3)

3
Ethernet Switch Operations

• Layer 2 Bridging and Switching Operations
• Bridges extend LAN distances, without some of the
  negative effects of repeaters and hubs
• Bridges were typically much more expensive than
  repeaters and hubs (were usually a PC running
  software to perform the bridging function)
• Bridges usually had only two interfaces, where
  hubs had multiple ports

4
Ethernet Switch Operations

• Layer 2 Bridging and Switching Operations
• Next major step in LAN devices was the LAN switch
• Does the same thing as a bridge
• Instead of using software, process could be done
  with a chip (sometimes called application-specific
  integrated circuits ASICs)
• Switches have more interfaces than bridges, are
  smaller, and do the same work faster
• As switch prices fell, bridges disappeared from
  the market

5
Ethernet Switch Operations

• The Forwarding and Filtering Decision
• Repeaters and hubs simply react to the incoming
  signal
• make no decisions and require no programming
  logic
• Receive, regenerate and send signal out all ports
  except the one on which it was received
• Bridges implemented their logic in software

6
Ethernet Switch Operations

• The Forwarding and Filtering Decision
• Switches implement their logic in hardware
• Run much faster than bridges
• Cisco makes switches that can forward hundreds of
  millions of Ethernet frames per second

7
Ethernet Switch Operations

• The Forwarding and Filtering Decision
• Filtering and forwarding logic
• Examine incoming signal interpret as 0s and 1s
  (OSI Layer 1 standards)
• Interpret the received bits based on Ethernet
  framing rules find MAC destination address in
  frame (OSI Layer 2 standards, IEEE 802.3 MAC
  sublayer)

8
Ethernet Switch Operations

• The Forwarding and Filtering Decision
• Filtering and forwarding logic (continued)
• Examine table that maps MAC addresses with
  corresponding interfaces
• Find table entry that matches the destination MAC
  address of frame
• If frame came in on a different interface than
  the one listed on the table, process is called
  forwarding the frame
• If the frame came in on the same interface as the
  one it was received on, discard it (this is
  called filtering)

9
Ethernet Switch Operations

• The Forwarding and Filtering Decision
• The table a bridge or switch refers to may be
  called
• Bridging table
• Switching table
• MAC address table
• Forwarding table
• Content Addressable Memory (CAM) table

10
Ethernet Switch Operations

• A Bridge Filtering Decision Based on the CAM

11
Ethernet Switch Operations

• A Bridge Forwarding Decision Based on the CAM

12
Ethernet Switch Operations

• Learning CAM Table Entries and Flooding Unknown
  Unicasts
• Switches and bridges learn entries in the CAM
  dynamically
• They use this logic
• Examine the source MAC address of the frame and
  the interface on which it was received
• Add that source MAC address and corresponding
  interface to the table

13
Ethernet Switch Operations

• Learning CAM Table Entries One Switch

14
Ethernet Switch Operations

• Learning CAM Table Entries Two Switches

15
Ethernet Switch Operations

• Handling Unknown Unicasts
• Switches typically learn CAM entries for all
  working devices on the LAN as soon as those
  devices start sending data
• Sometimes a switch receives a frame that does not
  have a CAM entry this is an unknown unicast
  frame
• The switch sends the unknown unicast frame out
  all ports, a process called flooding

16
Ethernet Switch Operations

• Forwarding Broadcasts and Multicasts
• Unicast frame has a destination MAC address of a
  single NIC or interface
• Broadcast frames are sent to a destination MAC
  address of FFFF.FFFF.FFFF.FFFF and are delivered
  to all devices on the LAN
• Multicast frames are sent to one of a range of
  MAC addresses

17
Ethernet Switch Operations

• Flooding Unknown Unicasts

18
Ethernet Switch Operations

• Forwarding Broadcasts and Multicasts
• Multicast addresses provide a way to send certain
  frames to a subset of devices
• Streaming video
• Some low-end switches flood multicasts like
  broadcasts
• Higher-end switches allow multicasting, making
  the process more efficient

19
Ethernet Switch Operations

• Different Forwarding Behavior for Multicasts

20
Ethernet Switch Operations

• The Cisco Switch CAM
• All switches and bridges use some table that
  lists the MAC address and port through which each
  MAC address can be reached
• Cisco calls this the CAM (Content Addressable
  Memory)
• The MAC address is input into the memory and CAM
  instantly outputs the table entry
• This process occurs quickly, every time,
  regardless of table size

21
Ethernet Switch Operations

• Switch Internal Processing
• The amount of time it takes for a frame to
  progress through a network from one device to
  another is called latency
• Some factors that affect latency cannot be
  improved, such as propagation delay (the amount
  of time it takes for electricity to go from one
  end of the network to another)
• Other types of delay vary with network
  conditions frames may be waiting in a buffer
  (queuing delay)

22
Ethernet Switch Operations

• Switch Internal Processing Factors that Impact
  Latency
• The finite speed that signals can travel
  (propagation delay)
• Circuit delays caused by electronics
• Software delays caused by software decisions
  being made
• Delays caused by frame contents and location of
  the frame switching decisions

23
Ethernet Switch Operations

• Store-and-Forward Switching
• Switch receives entire frame before forwarding it
• Advantages of store-and forward switching
• FCS field is at end of frame frame can be
  checked for an error
• Can check for rare error in which the 802.3
  Length field does not match the Data field length
• Can forward between ports running at different
  speeds (asymmetric switching)
• Disadvantage
• More latency than other switching types

24
Ethernet Switch Operations

• Cut-Through Switching
• Destination MAC address is located at beginning
  of Ethernet frame
• Advantage of cut-through switching
• Once destination MAC address is read, switch can
  begin forwarding frame
• Less latency than store-and-forward
• Disadvantages of cut-through switching
• Cannot check FCS may forward frames with errors
• Forwards before some legitimate collisions have
  occurred
• Only works with symmetric switching

25
Ethernet Switch Operations

• Fragment-Free Switching
• Overcomes a problem that cut-through switching
  has cut-through is too fast
• Collisions should occur while a frames first 64
  bytes are being transmitted
• Cut-through switching often begins transmitting
  before 64 bytes are received
• Cut-through switching can forward collision
  fragments

26
Ethernet Switch Operations

• Fragment-Free Switching
• Fragment-free switching waits until it has
  received first 64 bytes to begin transmitting
• Ensures switch does not forward frames that have
  collided

27
Ethernet Switch Operations

• Cisco Enterprise Switch Internal Processing
  Paths

28
Ethernet Switch Operations

• Spanning Tree Protocol
• Most LAN design include redundant physical paths
• A trunk is a link between two switches sometimes
  called a backbone link
• Spanning tree protocol (STP) prevents switching
  loops from the logic used to forward unknown
  unicast and broadcast frames

29
Ethernet Switch Operations

• Typical Enterprise Campus Building Block Design,
  with Redundancy

30
Ethernet Switch Operations

• The Problem That STP Solves Switching Loops

31
Ethernet Switch Operations

• The Problem That STP Solves Switching Loops
• In previous slide, if PC1 sends a broadcast, it
  goes around LAN in both directions
• Each switch broadcast the frame(s) out every port
  (except the one on which it was received)
• This process continues for a long time,
  continuing until no other traffic can be sent
  over the LAN a broadcast storm

32
Ethernet Switch Operations

• STP Protocol STP Blocking
• STP makes some ports quit forwarding or receiving
  frames
• An interface that is not allowed to process
  traffic by STP is considered to be in an STP
  blocking state
• In the figure that follows, SW3s port 1 is in a
  blocking state it receives the broadcast frame
  but ignores it
• STP causes the LAN to use particular paths and
  leaves others idle and unused

33
Ethernet Switch Operations

• IEEE 802.1D STP Interface States

34
Ethernet Switch Operations

• IEEE 802.1D STP Interface States
• The forwarding and blocking states are the most
  common, because a working network interface
  stabilizes into one of these states
• Failed interfaces stabilize into a disabled state
• Listening and learning states are used to solve
  problems with CAM tables

35
Ethernet Switch Operations

• Stable STP Topology and Switch CAMs in a
  Three-Switch Network

36
Ethernet Switch Operations

• Changing the CAM with the Listening and Learning
  States
• The topology can fail when a trunk fails or when
  a new trunk comes up
• STP determines the topology by having switches
  send bridge protocol data units (BPDUs) to each
  other
• BPDUs and the Spanning Tree Algorithm (STA) are
  part of the IEEE 802.1D standard
• Information learned allows the switches to
  determine the topology and decide which
  interfaces should forward and which should block
  frames

37
Ethernet Switch Operations

• Changing the CAM with the Listening and Learning
  States
• The listening and learning states are used by STP
  when it needs to transition to a new topology
• An STP topology refers to the topology of the
  network when each interface is in one of three
  stable states
• STP remains in the stable topology until
  something happens
• A trunk goes down (perhaps cut)
• The network engineer shuts down a trunk
• A new switch is added
• An interface fails

38
Ethernet Switch Operations

• Changing the CAM with the Listening and Learning
  States
• Switches use listening and learning states as
  interim states when transitioning an interface
  for two reasons
• For the switches CAM table entries to time out
  (during the listening state)
• For the switches to relearn the MAC addresses and
  (possibly different) interfaces used to reach the
  MAC addresses

39
Ethernet Switch Operations

• A New STP Topology After a Failure

40
LAN Design Collision Domains and Broadcast
Domains

• Collision Domains
• A collision domain is a set of LAN interfaces for
  which a frame sent out any two of these
  interfaces, at the same time, would cause a
  collision
• Hubs repeat signals out interfaces and do not
  consider CSMA/CD logic, so any frames sent
  simultaneously will collide
• The terms shared bandwidth and shared media refer
  to the fact that the devices in a hubbed network
  share the same media and bandwidth

41
LAN Design Collision Domains and Broadcast
Domains

• One Collision Domain with One 10BASE-T Hub

42
LAN Design Collision Domains and Broadcast
Domains

• Large/Long Collision Domains
• The 5-4-3 (or 5-4-3-2-1) Rule for 10BASE-T
  networks
• 5 segments of network media
• 4 repeaters or hubs at most
• 3 links at most, between two end-user devices
• If 5 segments exist between two end-user devices,
  2 segments must not have any end-user devices
  connected to them
• Its all 1 large collision domain

43
LAN Design Collision Domains and Broadcast
Domains

• One Collision Domain with Multiple 10BASE-T Hubs

44
LAN Design Collision Domains and Broadcast
Domains

• Large/Long Collision Domains
• The 5-4-3-2-1 rule for 10BASE-T restrictions are
  required due to the round-trip time of the
  collision domain
• Within one collision domain, all the devices
  share the 10 Mbps of bandwidth
• Within one collision domain, a (practically)
  simultaneous transmission of a frame by two or
  more PCs results in a collision

45
LAN Design Collision Domains and Broadcast
Domains

• Large/Long Collision Domains
• The more PCs in a collision domain, the less
  efficient it is
• The more frames, the more collisions
• The more collisions, the more time sent waiting
  to resend frames
• Once a LAN reaches about 30-40 of bandwidth
  utilization, the number of collisions increases
  dramatically

46
LAN Design Collision Domains and Broadcast
Domains

• High LAN Utilization Resulting in Much Higher
  Percentage of Collisions

47
LAN Design Collision Domains and Broadcast
Domains

• Large/Long Collision Domains
• Large collision domains should not be used for
  the following reasons
• Shared bandwidth as the size of the collision
  domain grows, each device has less available
  bandwidth
• Higher utilization the more devices in a single
  collision domain, the better the chance of a
  collision and of driving the utilization rate
  higher

48
LAN Design Collision Domains and Broadcast
Domains

• Creating Many Small Collision Domains
• Significantly reduces the negative effects of a
  large collision domain
• Process of breaking a LAN into multiple collision
  domains is called segmentation
• Switches, bridges, and routers can segment LANs
  into multiple collision domains

49
LAN Design Collision Domains and Broadcast
Domains

• Two LANs with Many Small Collision Domains

50
LAN Design Collision Domains and Broadcast
Domains

• Creating Many Small Collision Domains
• Benefits of segmenting 10BASE-T LANs
• Design rules (5-4-3-2-1) apply to each individual
  collision domain
• With smaller collision domains, reaching the
  point of utilization where performance is
  degraded is less likely
• Each domain gets its own bandwidth, so fewer
  devices are sharing the available bandwidth

51
LAN Design Collision Domains and Broadcast
Domains

• Creating Many Small Collision Domains
• When switches are used on the LAN, the terms
  switched LAN and switched bandwidth are used
• Each switch port connects to a separate collision
  domain
• Connecting a single end-user device to each
  switch port is a process called microsegmentation

52
LAN Design Collision Domains and Broadcast
Domains

• Creating Many Small Collision Domains
• Microsegments meet the requirements to allow full
  duplex
• Full duplex gives twice the bandwidth
• A 24 port 10BASE-T hub shares 10 Mbps of
  bandwidth among 24 ports
• A 24 port 10BASE-T switch gives each port 20 Mbps
  of bandwidth

53
LAN Design Collision Domains and Broadcast
Domains

• Main Benefits of Using Many Small Collision
  Domains
• Collision domain design rules are easier to
  achieve
• Smaller collision domains reduce the probability
  of LAN overutilization
• Each collision domain gets its own separate
  switched bandwidth
• With a collision domain consisting of only two
  interfaces/NICs, full duplex can be used

54
LAN Design Collision Domains and Broadcast
Domains

• How Switches and Bridges Prevent Collisions
• Switches reduce or prevent collisions by
  buffering or queuing frames
• Repeaters and hubs do not perform buffering
• Bridges, switches and routers follow CSMA/CD
  rules if not using full duplex

55
LAN Design Collision Domains and Broadcast
Domains

• Switch Buffering Example

56
LAN Design Collision Domains and Broadcast
Domains

• Layer 2 Broadcast Domains
• A broadcast domain is
• The set of LAN interfaces (including NICs and
  network device interfaces) for which a broadcast
  frame sent by one device with be forwarded to all
  other interfaces in that same broadcast domain
• Bridges and switches forward broadcasts
• Routers do not forward broadcasts

57
LAN Design Collision Domains and Broadcast
Domains

• One Router Creating Two Broadcast Domains

58
LAN Design Collision Domains and Broadcast
Domains

• Performance Impact of Multicast and Broadcast
  Domains
• PC NICs see all frames on the LAN
• PC NICs can ignore unicast frames not for them
• PC NICs must send multicast and broadcast frames
  to their CPU for processing, which affects PC
  performance
• This is less of an issue today with fewer
  proprietary network protocols doing broadcasts
  and with more powerful processors

59
LAN Design Collision Domains and Broadcast
Domains

• NIC Giving Broadcasts and Multicasts to the CPU

60
LAN Design Collision Domains and Broadcast
Domains

• More Broadcasts, Less CPU Capacity for End-User
  Work

61
LAN Design Collision Domains and Broadcast
Domains

• The Impact of Broadcasts and Multicasts Today
• Biggest risk is in wasting CPU cycles from
  multicasts
• Switches flood multicasts just like broadcasts
• LAN engineers must enable multicast optimization
  tools in switches to prevent switches from
  flooding multicasts to every device in the LAN

62
LAN Design Collision Domains and Broadcast
Domains

• The Impact of Broadcasts and Multicasts Today
• Broadcasts such as RIP and ARP dont cause
  problems in todays networks, but did in the past
  when networks were slower
• ARP remembers the info it learns, so an
  individual PC might not send one ARP per minute
• RIP broadcasts may be sent by routers and UNIX
  workstations now most UNIX workstations have it
  turned off by default so these are no longer an
  issue

63
LAN Design Collision Domains and Broadcast
Domains

• Identifying Networking Devices by OSI Layer
• Repeaters and hubs are Layer 1 devices
• Bridges and switches are Layer 2 devices
• Routers are Layer 3 devices

64
LAN Design Collision Domains and Broadcast
Domains

• Sample Network with Collision Domains and
  Broadcast Domains Shown

65
LAN Design Collision Domains and Broadcast
Domains

• Data Flow with Layer 1, Layer 2, and Layer 3
  Devices

66
LAN Design Collision Domains and Broadcast
Domains

• The Ambiguous Term Segment
• Three main uses of the term segment
• LAN concepts a segment is a collision domain
• LAN (physical) in a LAN using a bus topology, a
  segment is a continuous electrical circuit, often
  connected to other segments with repeaters
• TCP the process of taking a large piece of data
  and breaking it into smaller pieces one of those
  pieces

67
Summary

• Bridges and switches work the same way regarding
  basic forwarding, learning, flooding and STP
• They build forwarding tables by examining the
  source MAC addresses of incoming frames
• They make filtering and forwarding decisions by
  looking at the destination MAC address of the
  frame and comparing it to the table
• They flood broadcast frames and also flood
  multicast frames, unless optimization features
  have been enabled

68
Summary

• Switches differ from bridges
• They have much more powerful hardware
• They use content addressable memory (CAM) to hold
  the switching table
• The CAM allows the switch to find a MAC address
  and its associated port very quickly every time
• Latency is the time that passes as a frame or
  packet is sent through the network
• Propagation delay is the time it takes for
  electrical or optical energy to pass over the
  cable, and contributes to latency

69
Summary

• Three internal switch processing options
• Cut-through switching begins forwarding the frame
  as soon as the destination MAC address is read
  does not check FCS to determine if frame is good
  low latency
• Store-and-forward switching receives the entire
  frame does error-checking necessary for
  asymmetrical switching
• Fragment-free switching waits for the first 64
  bytes to be received before beginning forwarding
  enables it to detect normal collisions

70
Summary

• Switches and bridges use Spanning Tree Protocol
  (STP) to identify and block redundant paths
  through the network gives a logical path with no
  loops
• A collision domain with a single device connected
  to a switch port is called a microsegment
• Microsegments use UTP cabling, allow the use of
  full duplex
• With no collisions possible, CSMA/CD can be
  disabled

71
Summary

• Placing a large number of PCs in a collision
  domain increases demand for bandwidth
• This increases possibility of collisions
• Breaking large collision domains into multiple
  smaller collision domains reduces the chance of
  collisions while adding bandwidth
• Separating LANs into more segments by using
  bridges and switches creates additional collision
  domains, one per bridge and switch port
• Broadcast domains are a set of devices in which
  if one device sends a broadcast, all other
  devices receive the broadcast Layer 3 devices
  (routers) separate broadcast domains