Chapter 4 - Network Layer (Data Plane)

Question 1

What does a sender do in the Network Layer

Answer 1

Encapsulates segments into datagrams and passes this to the link layer

Question 2

What do routers do?

Answer 2

• Examines header fields for all IP datagrams
• moves datagrams from input ports to output ports

Question 3

What are the two network-layer functions?

Answer 3

Forwarding: move packets from routers input link to the appropriate output link

Routing: Determines the route taken by packets from source to destination (routing algorithms)

Question 4

Data Plane vs Control Plane

Answer 4

Data plane: (per router) determines how datagram on router input port is forwarded to router output port

Control Plane: (Network wide logic) determines how datagram is routed from souce router to destination router

Question 5

Per router control plane?

Answer 5

Individual routing algorithm components in each and every router interact in the control plane

Question 6

Software-Defined Networking (SDN) control plane

Answer 6

Remote controller computes, installs tables in routers

Question 7

Example network service models for INDIVIDUAL datagrams

Answer 7

• guaranteed delivery
• guaranteed delivery with less than 40 msec delay

Question 8

Example services for a FLOW of datagrams

Answer 8

• in order datagram delivery
• guaranteed minimum bandwidth to flow
• restrictions on changes in inter-packet spacing

Question 9

Best effort service model

Answer 9

Bandwidth - none
Loss - no
Order - no
Timing - no

Question 10

Why Best effort model?

Answer 10

• Simplicity
• sufficient provisioning of bandwidth allows performance to be “good enough”
• replicated, application-layer distributed services
• congestion control of “elastic” services

Question 11

Router Architecture

Answer 11

• Routing processor
• High speed switching fabric
• Input ports
• Output ports

Question 12

Input Port Functions

Answer 12

Physical Layer: bit-level reception

Link layer protocols

Decentralized switching:

Question 13

What is decentralized switching

Answer 13

• using the header field values, lookup the output port from the forwarding table in the input port memory
• the goal is to complete the input port processing at “line speed”
• if the datagrams arrive faster than the forwarding rate into switch fabric, then they will be queued.

Question 14

What is longest prefix matching

Answer 14

When looking for a forwarding table entry for given destination address, you use the longest address prefix that matches the destination address.

Question 15

What do switching fabrics do?

Answer 15

• transfer the packet from the input port to the appropriate output port

Question 16

What is the switching rate?

Answer 16

rate at which packets can be transferred from inputs to outputs

Question 17

Switching via memory

Answer 17

• packets are copied to the system’s memory
• the speed is limited by the memory bandwidth (2 bus crossings per datagram)

Question 18

Switching via bus

Answer 18

• datagram from the input port memory is transferred to the output port memory via a SHARED BUS
• Bus contention: the switching speed is limited by bus bandwidth

Question 19

Switching via interconnection network

Answer 19

-Crossbar: interconnected busses

• multistage switch: nxn switch from multiple stages of smaller switches
• exploiting parallelism: fragment datagram into fized length cells on entry. Switch cells through the fabric, reassemble datagram at exit

Question 20

What is input port queuing?

Answer 20

If the switch fabric is slower than the input ports combined, the the queuing may occur at input queues

Question 21

What is head-of-the-line (HOL) blocking?

Answer 21

a queued datagram at the front of the queue prevents others in the queue from moving foward.

Question 22

What are the two types of output port queuing?

Answer 22

• buffering
• scheduling dscipline

Question 23

What is buffering?

Answer 23

• required when datagrams arrive from fabric faster than link transmission rate.
• datagrams can be lost due to congestion, lack of buffers

Question 24

What is scheduling discipline

Answer 24

chooses among the queued datagrams for transmission

• through PRIORITY SCHEDULING, you can determine who gets the best performance.

Question 25

What are the two types of buffer management?

Answer 25

- dropping packets when buffers are full - marking packets to signal congestion

Question 26

what is tail dropping?

Answer 26

dropping the arriving packet

Question 27

what is the priority dropping

Answer 27

drop or remove packets based on a priority basis

Question 28

What are the 4 types of packet scheduling

Answer 28

- first come first serve aka First in first out - priority - round robin - weighted fair queuing

Question 29

What is First in first out packet scheduling

Answer 29

- which ever packets arrive first, those will be processed first followed by subsequent packets

Question 30

what is priority packet scheduling

Answer 30

- packets are defined to be high priority or low priority - packets that are high priority are processed first, followed by low priority packets

Question 31

what is round robin packet scheduling?

Answer 31

-packets are divided into classes - the server will cycle through the classes, processing a packet from each class each time. - will also go to high priority classes first

Question 32

What is weighted fair queuing packet scheduling?

Answer 32

- generalized round robin - each class has a weight and gets a weighted amount of service in each cycle: w_i/sum(w_j) - minimum bandwidth guarantee (per traffic class)

Question 33

What are path -selection algorithms?

Answer 33

-routing protocols - SDN controller

Question 34

IP protocols?

Answer 34

-datagram format - addressing - packet handling conventions

Question 35

ICMP protocol

Answer 35

- error reporting - router "signaling"

Question 36

What is an IP address?

Answer 36

32 bit identifier associated with each host or router interface

Question 37

What is an interface?

Answer 37

- connection between host/router and physical link -routers typically have multiple interfaces - host typically has one or two interfaces

Question 38

What is a subnet?

Answer 38

- device interfaces that can physically reach each other WITHOUT PASSING THOUGH AN INTERVENING ROUTER

Question 39

What are the parts of an IP address?

Answer 39

subnet part: devices in same subnet have common high order bits host part: the remaining low order bits

Question 40

What is CIDR?

Answer 40

- subnet portion of address or arbitrary length - a.b.c.d/x where c is the # bits in subnet portion of address.