Internet Routing

Basics
 Back to basics J

Application
Application Application
Presentation (HTTP, DNS, FTP) Data (HTTP, DNS, FTP)
Session
E2E connectivity (app-to-app)
Transport Transport Transport
Transport – P o rt n u m b e rs (s o c k e ts ) Header Data
(TCP/UDP) (TCP/UDP)

Device to device
Network Internet IP Transport Internet
– IP v 4 /IP v 6 a d d re s s Header Header Data
(IPv4/IPv6) (IPv4/IPv6)

Data Link Media access Frame IP Transport

Netw ork control Data Netw ork
Header Header Header
Access Access
Physical – M A C a d d re s s
(Ethernet, PPP) (Ethernet, PPP)
0011010100000111

Addressing is the key!

2
 Internet/Network Layer

• Host to host communication across networks

– Addressing
• unique and hierarchical network-wide address
– Routing
• the best path to the destination

• Current protocols
– IPv4 and IPv6

3
 L3 Device/Router
• L3 device gets the packet one step closer
– The next hop to reach the destination!

• Router
– Exchanges network information

– Finds the best path to a destination, and

– Forwards the packet to the next hop (a step closer) to reach

the destination

4
 Best path lookup – Routing
Decision

• Inspects the destination address of the packet

– Network portion

• Looks up its routing table for a “best match”

– Longest matching left-most bits

• If no match, checks for default route

– If no default route, drop the packet!

5
 Best path (route) lookup

Dest IP: 2001:db8:1::1/128

2001:db8::/32
R3

GE 1/0
GE 0/0
R1 R2 GE 1/1

R2#sh ipv6 route

2001:db8:1::/48
R4
2001:db8::/32 via R3
2001:db8:1::/48 via R4
………………………

2001:db8::/32 0010000000000001:1101101110000000::

2001:db8:1::/48 0010000000000001:1101101110000000:0000000000000001::

6
 Best path (route) lookup

Dest IP: 2001:db8:1::1/128

2001:db8::/32
R3

GE 1/0
GE 0/0
R1 R2 GE 1/1

R2#sh ipv6 route

2001:db8:1::/48
R4
2001:db8::/32 via R3
2001:db8:1::/48 via R4
Match! ………………………

2001:db8:1::1 0010000000000001:1101101110000000:0000000000000001:0:0:0:0:0000000000000001

FFFF:FFFF:: 1111111111111111:1111111111111111:0000000000000000:0:0:0:0:0000000000000000
AND
(/32)

2001:db8:: 0010000000000001:1101101110000000::

7
 Best path (route) lookup

Dest IP: 2001:db8:1::1/128

2001:db8::/32
R3

GE 1/0
GE 0/0
R1 R2 GE 1/1

R2#sh ipv6 route

2001:db8:1::/48
R4
2001:db8::/32 via R3
2001:db8:1::/48 via R4

Match! ………………………

2001:db8:1::1 0010000000000001:1101101110000000:0000000000000001:0:0:0:0:0000000000000001

FFFF:FFFF:FFFF:: 1111111111111111:1111111111111111:1111111111111111:0:0:0:0:0000000000000000
AND
(/48)

2001:db8:1:: 0010000000000001:1101101110000000:0000000000000001::

8
 Best path (route) lookup

Dest IP: 2001:db8:1::1/128

2001:db8::/32
R3

GE 1/0
GE 0/0
R1 R2 GE 1/1

R2#sh ipv6 route

2001:db8:1::/48
R4
2001:db8::/32 via R3
2001:db8:1::/48 via R4
………………………
Longest Match!
2001:db8:1::1 0010000000000001:1101101110000000:0000000000000001:0:0:0:0:0000000000000001

FFFF:FFFF:FFFF:: 1111111111111111:1111111111111111:1111111111111111:0:0:0:0:0000000000000000
AND
(/48)

2001:db8:1:: 0010000000000001:1101101110000000:0000000000000001::

9
 Packet Forwarding
• If a best match is found, the router determines
– the correct exit interface to reach the next-hop/destination

Is the best m atch Directly YES Forward to host on

connected
a subnet of … . local subnet
interface?

NO
YES Forward out the
Remote exit interface to
Network?
the next-hop

NO
NO Is there a YES Forward out the
Drop the gateway of exit interface to
packet! last resort? the next-hop

10
Process vs Hardware Switching
• Incoming packet forwarded to the control plane
(CPU)
– routing table (RIB) lookup, frame re-write (next-hop MAC),
and forwarded to the exit interface

Control Plane
RIB

Data Plane
Incoming Packets Outgoing Packets

11
Process vs Hardware Switching
• Instead two hardware based tables
– FIB derived from the RIB- all destinations and next-hops
– Adjacency table from the ARP table- L2 header info for each
next-hop in the FIB

Control Plane
RIB

Data Plane Outgoing Packets

Incoming Packets
FIB & Adjacency Table

12
 Internet Routing
• How does a user in VN access a service hosted in
the BT?

– The ISP in VN could directly connect to the ISP in BT

• Neither scalable nor economical

– Instead, the VN ISP shares its network information with its

neighbor ISPs

– The ISP in BT does the same with its own neighbors

– Neighbor ISPs propagate the information to their neighbors,

and so on…
• Eventually, they both learn about each other’s network!

13
 Internet Routing

VN BT
AS-X AS-N
Routing flow
Traffic flow

SG IN
AS-Y AS-M

Exchange of network information – Routing

Networks (ASes) connected together – Internet

14
 Autonomous System (AS)
• A group of networks with the same routing policy
(external)
– Usually under single administrative control

AS-X

15
 Routing Flow & Traffic Flow
• Traffic and network info always flow in opposite
direction!

– network info exchanged in both directions for bi-directional

traffic flow
Packet Flow

Receive Routing Flow

Advertise
AS X AS Y
R1 Accept R2
Routing Flow
Send

AS X Packet Flow

16
Routing & Traffic Flow: Internet
VN BT
AS-1 Routing flow AS-7
Traffic flow

SG IN
AS-3 AS-5

• For user (N1) in AS1 to send traffic to user (N7) in

AS7:
– AS7 must originate and announce N7 to AS5.
– AS5 must accept N7 from AS7, and advertise to AS3.
– AS3 must accept and forward N7 to AS1
– AS1 must accept N7 from AS3

17
 Routing Policy
• To manipulate/control traffic flow in/out of a
network

– manipulate inbound routing info to influence outgoing traffic

– manipulate outbound routing info to influence incoming

traffic

18
 Routing Protocols
• How do routers exchange network information with
each other?
– Routing Protocols!
– IGP & EGP

19
 Interior Gateway Protocol
(IGP)

• To exchange network info within an AS

– Allows all routers within an AS to learn about each other
– To carry infrastructure information (loopbacks & ptp)
• No customer routes!
– The design goal is scalability and fast convergence
• Hence, minimise the number of prefixes carried in IGP!

• Two most widely used IGPs in operator networks

– OSPF & IS-IS
• Uses the SPF algorithm
• Best path selection based on lowest cost/metric
• Supports hierarchical routing – scalability!

20
 Exterior Gateway Protocol
(EGP - BGP)

• To exchange network information between ASes

– Implement routing policies (manipulate traffic path)
– Define administrative boundary

• BGP is the de facto EGP!

21
Routing Protocols Hierarchy
Other ISPs
eBGP

iBGP &
OSPF/IS-IS

eBGP Static/eBGP
IX or direct Customers
Peers

22
 How it all works

eBGP eBGP

iBGP iBGP iBGP

IGP IGP IGP

AS X AS Y AS Z

Barry Greene & Philip Smith “Cisco ISP Essentials”

23
24