cps-356- computer networks class 8: ip forwarding+ routing theophilus benson based partly on lecture...

CPS-356- Computer NetworksClass 8: IP Forwarding+ Routing

Theophilus Benson

Based partly on lecture notes by Rodrigo Fonseca, David Mazières, Phil Levis, John Jannotti

Admini-strivia

• Midterm 1:– Day after UNC game: New proposed dates:• 02/24/2015

• HW #1: Going Up Tomorrow on Website (due in a week 02/12/2015)

Today’s Lecture

• Forwarding– IP-Address/IP-Packet Format– Fragmentation– Debugging the network: ICMP– Getting IP-Address: ARP + DHCP

• Routing– Intra-Domain Routing– Distance Vector Protocol• Loop Detection + Avoidance

Format of IP Addresses

Classed Addresses• Pros

– Very simple to use and implement– Allows for hierarchical routing– Use first 3 bits to determine

addresses class (A, B, C)– Based on class you know what

bits to ignore

• Cons– Wasteful allocation– Statically specify network and

host portion of address

CIDR Addresses• Pros

– Efficient allocation of resources

– dynamically specify network and host portion of address

• Cons– More complex to implement

in hardware

Format of IP AddressesClassed Addresses (Static partitioning of Network/host portions)

• Class A (8-bit prefix), B (16-bit), C (24-bit)

CIDR(Dynamic partitioning of Network/hosts portions)

128.23.92.12

10000000

128.23.16.12/31

Specifies the prefix size: the number of bits in the network portion (NetMask)

11111111.11111111.11111111.11111110

10000000.00010111.00010000.00001100

Prefix size = 31 bitsHost size = 1 bit32-31=1Only 2^1 hosts in the network

Other CIDR Examples

128.23.16.12/32

11111111.11111111.11111111.00000000

10000000.00010111.00010000.00001100

128.23.16.12/24

11111111.11111111.11111111.11111111

10000000.00010111.00010000.00001100

Prefix size = 32 bitsHost size = 0 bit32-32=0Only 2^0 hosts in the network

Prefix size = 24 bitsHost size = 8 bits32 – 24 = 8Only 2^8 hosts in the network

Where Does IP-Address Fit Into a packet?

Ethernet

IP

Destination MAC AddressSource MAC Address

Length

Destination IP AddressSource IP Address

Type

Options Padding

TTL Protocol Hdr checksum

V Total Length

FragV TOS

Identification MMM

Src PortSeq Number

Offset Reserved

Dst Port

Ack NumberWindow

Data (Payload)

IP v4 packet format

Destination IP Address

Source IP Address

Options

Total LengthTOS

Identification

Hdr lenvers

TTL Protocol Hdr Checksum

Padding

Fragment Offset

• Forward based on destination address

IP v4 packet format


Source IP Address

Options

Total LengthTOS

Identification

Hdr lenvers


Padding

Fragment Offset


• TTL = Time to Live• Prevents forwarding loops• Decremented at each hop

IP v4 packet format


Source IP Address

Options

Total LengthTOS

Identification

Hdr lenvers


Padding

Fragment Offset



• Cut large packets into smaller ones• E.g. from Ethernet to ATM• From 1500B to 64B• MF: more fragments• DF: don’t fragment (return

an error to the sender)

MFDF

IP v4 packet format


Source IP Address

Options

Total LengthTOS

Identification

Hdr lenvers


Padding

Fragment Offset




an error to the sender)• Version = IPv4 or IPv6

MFDF

IP v4 packet format


Source IP Address

Options

Total LengthTOS

Identification

Hdr lenvers


Padding

Fragment Offset




an error to the sender)• Version = IPv4 or IPv6• Protocol = TCP/UDP?

IP v4 packet format


Source IP Address

Options

Total LengthTOS

Identification

Hdr lenvers


Padding

Fragment Offset




an error to the sender)• Version = IPv4 or IPv6• Protocol = TCP/UDP?

Header length == size of the header, which can vary because you can have an arbitrary number of optionsTotal length == length of header + payload

Today’s Lecture

• Forwarding– IP-Address/IP-Packet Format– Fragmentation– Network Error Messages (Debugging): ICMP– Getting IP-Address: ARP + DHCP

• Routing– Intra-Domain Routing: RIP

Why Do you need to Fragment Packets?

• Different networks have different MTUs.– Router may need to fragment packets to allow

them to cross different mediums

Le Theo Net (ATM)

DukeNet(Ethernet)

ATT(Ethernet)

MTU=1500 MTU=1500MTU=64

Implication of Fragmentation

• If a fragment is lost, must retransmit the whole packet!!! Why?

• Fragmentation delays reassembly of packet until all fragments are received

• Some people avoid fragmentation!!!!

What do Fragmented Packets look like?

• Use ‘identification’, ‘fragment offset’ and ‘MF’ bit in IP header– Set the ‘MF’ bit– Use the same ‘Id’ for all fragments– Offset present position in original packet

1400 Bytes

Rest of header

Start of header

0213 0 512 bytes

Rest of header

Start of header

0213 1

512 bytes

Rest of header

Start of header

64213 1

376 bytes

Rest of header

Start of header

128213 0

Internet Control Message Protocol (ICMP)

• Echo (ping)• Redirect• Destination unreachable (protocol, port, or host)• TTL exceeded• Checksum failed• Reassembly failed• Can’t fragment• Many ICMP messages include part of packet that

triggered them• See

http://www.iana.org/assignments/icmp-parameters

http://www.iana.org/assignments/icmp-parameters

ICMP message format

Example: Time Exceeded

• Code usually 0 (TTL exceeded in transit)• Discussion: traceroute

Example: Can’t Fragment

• Sent if DF=1 and packet length > MTU• What can you use this for?• Path MTU Discovery– Can do binary search on packet sizes– But better: base algorithm on most common

MTUs

Today’s Lecture



How do you Make a Packet

Ethernet

IP


Length


Source IP Address

Type

Options Padding


V Total Length

FragV TOS

Identification MMM

Src PortSeq Number

Offset Reserved

Dst Port

Ack NumberWindow

Data (Payload)

Comes with your hardware

???????

DNS gives this to you

Obtaining Host IP Addresses - DHCP

• Address must be assigned to each host by his network.– Manually: Tedious and error-prone: – Automatically: Dynamic Host Configuration Protocol• Client: DHCP Discover to 255.255.255.255 (broadcast)• Server(s): DHCP Offer to 255.255.255.255 (why

broadcast?)• Client: choose offer, DHCP Request (broadcast, why?)• Server: DHCP ACK (again broadcast)

• Result: IP-address, gateway, netmask, DNS server


Ethernet

IP


Length


Source IP Address

Type

Options Padding


V Total Length

FragV TOS

Identification MMM

Src PortSeq Number

Offset Reserved

Dst Port

Ack NumberWindow

Data (Payload)


???????


DHCP

What is the Destination Address?

• If dest. is in your network (e.g. Alice to Bob)– Then use the Destination’s Ethernet address.

• If dest. is not in your network (e.g Alice to Google)– Then use the gateway router’s Ethernet address.– The destination may use a different protocol

Le Theo Net (ATM)

DukeNet(Ethernet)

Alice

Bob

Google

Ethernet

Ethernet

ATMATMEthernet

How do you find this destination address?

• Check local ARP table– If found use it. (DONE!)– Start sending packets!

DukeNet(Ethernet)

Alice

Bob

Ethernet

Ethernet

ATMEthernet


• Check local ARP table– If found use it. (DONE!)

• Compare my IP with dest IP– In same network?

• Then ARP request for Dest IP

– In different Networks?• Then ARP request for Router IP

Alice: 128.23.16.12/30

Google: 128.16.16.16Bob: 128.23.16.14

DukeNet:128.23.16.12/30 4 addresses128.23.16.12– 128.23.16.16

Alice->Bob: same networkAlice->Google: diff network


Alice: 128.23.16.12/30

Google: 128.16.16.16Bob: 128.23.16.14

DukeNet:128.23.16.12/30 4 addresses128.23.16.12– 128.23.16.16

Alice->Bob: same networkAlice->Google: diff network

DukeNet(Ethernet)

Alice

Bob

Ethernet

Ethernet

ATMEthernet

How ARP works.

DukeNet(Ethernet)

Alice

Bob

Ethernet

Ethernet

ATMEthernet

I am:128.23.16.12Who is IP: 128.23.16.14

How ARP works.

DukeNet(Ethernet)

Alice

Bob

Ethernet

Ethernet

ATMEthernet

I am:128.23.16.12Who is IP: 128.23.16.14

Now I know who:128.23.16.12 is!


How ARP works.

DukeNet(Ethernet)

Alice

Bob

Ethernet

Ethernet

ATMEthernet


I am:128.23.16.14MacAdd: 02………..

Now I know who:

128.23.16.14 is!

ARP Ethernet frame format

• Why include source hardware address?


Ethernet

IP


Length


Source IP Address

Type

Options Padding


V Total Length

FragV TOS

Identification MMM

Src PortSeq Number

Offset Reserved

Dst Port

Ack NumberWindow

Data (Payload)


ARP


DHCP

Today’s Lecture



Routing• Routing is the process of updating forwarding

tables– Routers exchange messages about routers or

networks they can reach– Goal: find optimal route for every destination– … or maybe a good route, or any route (depending

on scale)• Challenges– Dynamic topology– Decentralized – Scale

Scaling Issues• Every router must be able to forward based on

any destination IP address– Given address, it needs to know next hop– Naïve: one entry per address– There would be 108 entries!

• Solutions– Hierarchy (many examples)– Address aggregation

• Address allocation is very important (should mirror topology)

– Default routes

IP Connectivity

• For each destination address, must either:– Have prefix mapped to next hop in forwarding table– Know “smarter router” – default for unknown

prefixes• Route using longest prefix match, default is

prefix 0.0.0.0/0• Core routers know everything – no default• Manage using notion of Autonomous System

(AS)

Internet structure, 1990

• Several independent organizations• Hierarchical structure with single

backbone

Internet structure, today

• Multiple backbones, more arbitrary structure

Autonomous Systems

• Correspond to an administrative domain– AS’s reflect organization of the Internet– E.g., DukeNet, large company, etc.– Identified by a 16-bit number

• AS are also called ISP– ISP = Internet Service Providers

Le Theo Net

DukeNet ATT

• AS’s choose their own local routing algorithm• How should A,B,C,D do routing?

• AS’s want to set policies about non-local routing• Should DukeNet use Link 1 or 2 to ATT?

• AS’s need not reveal internal topology of their network• That Duke Net has 4 routers

A

DC

B

Lnk 1

Lnk2

Inter and Intra-domain routing• Routing organized in two levels• Intra-domain routing

– Complete knowledge, strive for optimal paths– Scale to ~100 networks– Today

• Inter-domain routing– Aggregated knowledge, scale to Internet– Dominated by policy

• E.g., route through X, unless X is unavailable, then route through Y. Never route traffic from X to Y.

– Policies reflect business agreements, can get complex– Next lecture

Le Theo Net

DukeNet ATTA

DC

B

Lnk 1

Lnk2

Intradomain:Routing inside DukeNET

Interdomain:Routing across DukeNet, ATT, TheoNet

Today’s Lecture



Network as a graph

• Nodes are routers• Assign cost to each edge– Can be based on latency, b/w, queue length, …

• Problem: find lowest-cost path between nodes– Each node individually computes routes

Basic Algorithms

• Two classes of intra-domain routing algorithms• Distance Vector (Bellman-Ford SP Algorithm)– Requires only local state– Harder to debug– Can suffer from loops

• Link State (Djikstra-Prim SP Algorithm)– Each node has global view of the network– Simpler to debug– Requires global state

Distance Vector• Local routing algorithm• Each node maintains a set of triples– <Destination, Cost, NextHop>

• Exchange updates with neighbors– Periodically (seconds to minutes)– Whenever table changes (triggered update)

• Each update is a list of pairs– <Destination, Cost>

• Update local table if receive a “better” route– Smaller cost

• Refresh existing routes, delete if time out

DV Example

B only exchanges information with A

and C

DV ExampleB only exchanges information with A

and C

Destination Cost Next Hop

A 1 A

C 1 C

D infinity --

E infinity --

F infinity --

G infinity --

B’s routing table@ time = 0D, 1

A, 1

DV ExampleB only exchanges information with A

and C


A 1 A

C 1 C

D 2 C

E infinity --

F infinity --

G infinity --

B’s routing table@ time = 0

D, 1A, 1

Calculating the best path• Bellman-Ford equation• Let:

– Db(d) denote the current best distance from b to d– C(b,c) denote the cost of a link from a to b

• Then Db(d) = mind(Db(d) , c(b,c) + Dc(d))• Routing messages contain D• D is any additive metric

– e.g, number of hops, queue length, delay– log can convert multiplicative metric into an additive one

(e.g., probability of failure)

C’s updateD, 1A, 1


A 1 A

C 1 C

D infinity --

E infinite --

F infinite --

G infinite --

Db(d) = mind(infinity, 1 + 1)

Db(A) = minA(1, 1 + 1)

Calculating the best path• Bellman-Ford equation• Let:

– Db(d) denote the current best distance from b to d– C(b,c) denote the cost of a link from a to b

• Then Db(d) = mind(Db(d) , c(b,c) + Dc(d))• Routing messages contain D• D is any additive metric

– e.g, number of hops, queue length, delay– asdf

DV Example


A 1 A

C 1 C

D 2 C

E 2 A

F 2 A

G 3 A

B’s routing table

G, 1, G

• F-G fails• F sets distance to G to infinity, propagates• A sets distance to G to infinity• A receives periodic update from C with 2-hop path to G• A sets distance to G to 3 and propagates• F sets distance to G to 4, through A

G, ∞, -G, 4, A

Adapting to Failures

G, 2, F

G, 2, DG, 3, D

G, 3, A

G, 1, GG, ∞,-G, 3,C

G, 4, A

Count-to-Infinity

• Link from A to E fails• A advertises distance of infinity to E• B and C advertise a distance of 2 to E• B decides it can reach E in 3 hops through C• A decides it can reach E in 4 hops through B• C decides it can reach E in 5 hops through A, …• When does this stop?

Good news travels fast

A

B

C

4 1

10

1

• A decrease in link cost has to be fresh information• Network converges at most in O(diameter) steps

Bad news travels slowly

A

B

C

4 1

10• An increase in cost may cause confusion with old information, may form

loops• Consider routes to A• Initially, B:A,4,A; C:A,5,B• Then B:A,12,A, selects C as next hop -> B:A,6,C• C -> A,7,B; B -> A,8,C; C -> A,9,B; B -> A,10,C;• C finally chooses C:A,10,A, and B -> A,11,C!

A 4 A

C 1 C

A 5 B

B 1 BB 4 B

C 5 B


A

B

C

4 1

10

12

• An increase in cost may cause confusion with old information, may form loops

• Consider routes to A• Initially, B:A,4,A; C:A,5,B• Then B:A,12,A, selects C as next hop -> B:A,6,C• C -> A,7,B; B -> A,8,C; C -> A,9,B; B -> A,10,C;• C finally chooses C:A,10,A, and B -> A,11,C!

A 6 C

C 1 C

A 5 B

B 1 BB 11 C

C 10 C

A6C


A

B

C

4 1

10

12



A 7 C

C 1 C

A 6 B

B 1 BB 11 C

C 10 C


A

B

C

4 1

10

12



A 11 C

C 1 C

A 10 C

B 1 BB 11 C

C 10 C

How to avoid loops

• IP TTL field prevents a packet from living forever– Does not repair a loop

• Simple approach: consider a small cost n (e.g., 16) to be infinity– After n rounds decide node is unavailable– But rounds can be long, this takes time

• Problem: distance vector based only on local information


A

B

C

4 1

10

12

A 11 C

C 1 C

A 10 C

B 1 BB 11 C

C 10 C

• Why did it take a while to converge?

Better loop avoidance

• Split Horizon– When sending updates to node A, don’t include

routes you learned from A– Prevents B and C from sending cost 2 to A

• Split Horizon with Poison Reverse– Rather than not advertising routes learned from A,

explicitly include cost of ∞.– Faster to break out of loops, but increases

advertisement sizes

Warning

• Split horizon/split horizon with poison reverse only help between two nodes– Can still get loop with three nodes involved– Might need to delay advertising routes after

changes, but affects convergence time

Today’s Lecture

• Forwarding– IP-Address/IP-Packet Format– Fragmentation– Network Error Messages (Debugging): ICMP– Getting IP-Address: ARP + DHCP


• Next class:– Intra-Domain Routing: OSPF, OSPF v RIP

cps-356- computer networks class 8: ip forwarding+ routing theophilus benson based partly on lecture...

Documents