Term
What is the responsability of the Ethernet as the Link layer? |
|
Definition
- Receive/transmit datagrams from the network layer
- Encapsulate these as the payload of an Ethernet frame
- Detect transmission errors
- Use acknowledgements in the case of 802.11
- Implies IP layer can determine Ethernet addresses
- Need a method to resolve an IP address to mac address
|
|
|
Term
What are the elements of the Link Layer? |
|
Definition
- Logical Link Control
- Provides the interface to layer 3 (IP layer)
- Supports error detection, ARQ pipelining
- May use ACKs
- Not used in wired Ethernet
- Medium Access Control (MAC) protocol
- Controls access to the broadcast media
- Handles multiple host access to the same collision domain
- Shared Ethernet uses CSMA/CD
|
|
|
Term
What is the Logical Link Control (LLC)? |
|
Definition
Controls frame synchronization, flow control and error checking:
- Provides interface to the IP Layer
- Supports error detection , ARQ pipelining
- May use ACKs (not in wired Ethernet)
|
|
|
Term
What is the Medium Access Control (MAC) protocol? |
|
Definition
- Controls access to the broadcast media
- Handles mutliple host access to the same collision domain
|
|
|
Term
What are the characteristics of Ethernet Addresses? |
|
Definition
- Link Layer addresses
- Often referred as MAC addresses
- Currently defined to be 48 bits, extensible to 64 bits
- Needs to be unique
- 24 bits used for vendor allocations
- Last 24 bits assigned by vendor
|
|
|
Term
What are some of the components of the Ethernet frame? |
|
Definition
- Reminder - includes:
- 48-bit source and destination addresses
- 1 byte SFD to sync Tx and Rx for the start of frame
- 802.1Q tag for optional VLAN ID and frame priority
- Minimum data size 46 bytes
- Maximum set by link Maximum Transmission Unit (MTU)
|
|
|
Term
What are some types of ethernet transmission? |
|
Definition
- Generally uses block encoding, e.v. 8b/10b
- 10/100Gig Ethernet uses 64b/66b
|
|
|
Term
What are some principles of Ethernet reliability? |
|
Definition
- Wired Ethernet doesn't use acknowledgements
- It may however detect errors
- On error, the 'bad' frame is simply dropped
- Not passed up to receiving host's IP layer
- No negative acknowledgement (NACK) sent
- With no ACKs there is no inherent retransmission
- Recovery needs to be handled by higher layer
- By TCP which inherently provides retransmission
- By the application, if UDP, which does not
- Wired iternet is a reliable medium
|
|
|
Term
What is address resolution protocol (ARP)? |
|
Definition
- Maps an IP address to a physical machine address (MAC) that is recognized in the local network.
- A table is used to maintain a correlation between each MAC address and its corresponding IP address.
- If the current entry does not exist in the table, a broadcast message is sent within the link layer of the network asking for this person - and this person will respond with the required information
|
|
|
Term
How does the Address Resolution Protocol work? |
|
Definition
- Uses a link layer broadcast message
- Ethernet broadvast address is ff:ff:ff:ff:ff:ff
- Asks "Who has this IP address?"
- This is seen by all hosts in the same layer 2 network
- Host with the target IP address responds
|
|
|
Term
What are some of the flaws of ARP? |
|
Definition
- Potentially open to spoofing
- Middle-man attacks can respond as the IP address being requested
- Sleeping nodes
- Can use proxy ARP to answer for them
- Change of IP or MAC addresss
- ARP probe
- To detect IP address clashes
|
|
|
Term
|
Definition
A table which contains entries of IP and MAC addresses which are checked before sending a broadcast message |
|
|
Term
What are the classes of data link networks? |
|
Definition
- Broadcast
- Multiple hosts in one layer 2 broadcast domain
- Hosts in the same broadcast domain fall into the same LAN
- The layer 2 broadcast domain cotrols where ARP travels
- Point to point
- Typically router to router
- Only two devices in the broadcast domain and corresponding subnet
|
|
|
Term
What are some Ethernet topologies? |
|
Definition
- Bus (old)
- Hosts 'tap' into a single medium (cable)
- Star (current)
- Each host connects to a port in concentrator device
|
|
|
Term
|
Definition
Connection point for devices in a network with multiple ports - when a packet arrives at one port it is copied to all the other ports. |
|
|
Term
|
Definition
- Device that filters and forwards packets between LAN segments.
- Operate at data link and network layer.
- Learns ethernet addresses of host(s) seen on each switch port.
|
|
|
Term
What are some characteristics of Ethernet switching? |
|
Definition
- Commodity switching new at 1Gbit/s
- Internal switch fabric between switch ports
- One switch port connects each host
- No need for CSMA/CD (Unitil a researcher plugs a hub in)
- Can run Cat5 at 1Gbit/s to about 100m
- May have a higher speed uplink port or module
|
|
|
Term
What are the 3 key principles of Switched Ethernet networks? |
|
Definition
- How switches learn which ports to forward received frames on to, to minimise traffic propagation
- How loops in switch topologies can be handled
- How virtual layer 2 networks can be deployed
|
|
|
Term
What are the principles of MAC learning in Ethernet Switching? |
|
Definition
- Allows bridge or switch to only forward frames to ports which serve devices they are addressed to
- Uses an Ethernet (MAC) table per port
- Observes incoming source MAC addresses
- Stores observed MAC source addresses in port table
- Then forward frames to taht address to taht port, and that port only
- Hosts only see traffic for them, or any broadcast/multicast
- If MAC address is not in any table, switch must flood to all ports
- Purge MAC tables periodically
|
|
|
Term
How to avoid loops in switched networks? |
|
Definition
- Using Spanning Tree Algorithms
- Using redundant paths
|
|
|
Term
What does the Spanning Tree (802.1d) Algorithm for Switched networks consist of? |
|
Definition
It uses BPDU (Bridge Protocol Data Unit)
>Switches use these to determine the root and the best path to root<
- Lets you create physical loops without breaking network
- Refers to'bridges' which in today's terms are switches
- Algorithm decides(elects) a root bridge (lowest MAC)
- Determines least cost paths to the root
- Enables root (facing) ports on each switch
- Other ports with a path to the root are blocked
- Alternative paths may be enabled if topology changes
|
|
|
Term
What is the 802.1d protocol? |
|
Definition
|
|
Term
What is the 802.1Q protocol? |
|
Definition
|
|
Term
What are the principles of Virtual LANs (802.1Q)? |
|
Definition
- Used to create 'virtual' switched networks
- And therefore virtual broadcast domains
- Ethernet frame may include a VLAN identifier
- 12-bit value within 802.1Q tag, so 4,096 values
- Can then place switch ports in specific VLANs
- Can carry multiple VLANs over one uplink
- Avoid needs to physically re-cable
- Can place one switch port in a different VLAN without touching cabling
|
|
|
Term
What protocol is the 802.1p? |
|
Definition
|
|
Term
What does the Ethernet frame priority (802.1p) protocol consist of? |
|
Definition
- A specification for giving Layer 2 switches the ability to prioratize traffic (and perform dynamic multicast filtering)
- Supported by use of the 802.1Q tag
- Three bits of the tag are used for 802.1p priority
- 1 is low, 7 is high priority
- 0 is just treated as regular traffic
|
|
|
Term
What is the 802.11 protocol? |
|
Definition
|
|