1. Border Gateway Protocol (BGP)
Between Autonomous Systems: uses path vector routing.
This routing protocol uses TCP and port #179. Messages are encapsulated in TCP.
For between autonomous systems:
Distance Vector not good to use because do not always want to use smallest hop count
Link state not good - Internet is too large for this approach. Data base too large, shortest path algorithm calculations do not scale well
Source: TCP/IP Protocol Suite, by Fourouzan

2. Path Vector Routing
Routing table entries consist of destination network, next router, autonomous system path to reach destination
Example path vector routing table
Destination Network Next Router Path
N01 R07 AS5, AS19, AS21
N02 R10 AS14, AS8, AS7
N03 R2 AS4, AS9, AS6
N04 R3 AS1, AS2
Autonomous Boundary Routers advertise to their neighbors how to get to networks inside their own autonomous systems

4. BGP Types of Messages
Types are: OPEN, UPDATE, KEEPALIVE,NOTIFICATION
Packet Formats:
All BGP packets have same common header:
Marker 16 bytes
Length 16 bits
Type 8 bits
Marker - Fields for authentication
Length - Length of entire BGP message including common header shown here
Type - Identifies which of the four message types this is.

5. My autonomous system 16 bits
Open Message
Use this message type to open a connection (TCP connection used) with neighbor
Version 8 bits
My autonomous system 16 bits
Hold Time 16 bits
BGP Identifier 32 bits
Option Length
Variable Length Option
Type in common header = 1
Version - Version is 4 at present in the internet
My Autonomous System - Autonomous System Number
Hold Time - Max number of seconds before must receive update or a keep alive message or we consider other router dead
BGP ID - Four byte IP of router sending
Options

6. Unfeasible Routes Length 8 bits
Update Message
Used to remove destinations that were advertised previously, announce new route. BGP is allowed to remove several destinations but only to add one new in each update message.
Unfeasible Routes Length 8 bits
Withdrawn Routes is of variable length
Path Attribute Length 16 bits
Variable Length Path Attributes
Variable Length Network Layer Reachability Info
Unfeasible Routes Length - Length of next field
Withdrawn Routes - List of routes to remove
Path Attributes Length - Length of next field
Path Attributes - Info like ORIGIN: RIP, OSPF, etc generated route info;
AS_PATH: List of autonomous systems through which
we go to get to destination;Next Hop: Router where we should send packet
Network Layer Reachability Info - Network IP and number of bits in mask

7. Variable Length Error data
Keep Alive Message
The Routers exchange Keep Alive messages. (2 Minutes common)
The message itself is just a common BGP Header Message with Type = 3
Notification Message
Sent when error is detected or router wants to close the connection.
Message consists of error codes, error subcase, and error data.
Type = 4
Error Code 8 bits
Error Subcode 8 bits
Variable Length Error data

8. spawn telnet route-server.ip.att.net
Trying
Connected to route-server.cbbtier3.att.net.
Escape character is '^]'.
CCC
############## route-server.ip.att.net ###############
######### AT&T IP Services Route Monitor ###########
This router maintains peerings with customer-facing routers
throughout the AT&T IP Services Backbone:
Atlanta, GA Austin, TX
Cambridge, MA Chicago,IL
Dallas, TX Detroit, MI
Denver, CO Houston, TX
Los Angeles, CA New York, NY
Orlando,FL Philadelphia, PA
Phoenix, AZ San Diego, CA
San Francisco, CA St. Louis, MO
Seattle, WA Washington, DC
This router has the global routing table view from each of the above
routers, providing a glimpse to the Internet routing table from the
AT&T network's perspective.

9. route-server>Kerberos: No default realm defined for Kerberos!
term len 0
route-server>sh ip bgp sum
BGP router identifier , local AS number 65000
BGP table version is , main routing table version
network entries using bytes of memory
path entries using bytes of memory
37373 BGP path attribute entries using bytes of memory
33667 BGP AS-PATH entries using bytes of memory
5 BGP community entries using 136 bytes of memory
0 BGP route-map cache entries using 0 bytes of memory
0 BGP filter-list cache entries using 0 bytes of memory
BGP using total bytes of memory
Dampening enabled history paths, 994 dampened paths
received paths for inbound soft reconfiguration
BGP activity /17627 prefixes, / paths, scan interval 60 secs
Neighbor V AS MsgRcvd MsgSent TblVer InQ OutQ Up/Down State/PfxRcd
never Idle (Admin)
w1d
w1d
w1d
w1d
w1d
d20h
w1d
w1d
w1d
w1d
d19h

10. BGP table version is 1286649, local router ID is 10.1.2.5
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,
S Stale
Origin codes: i - IGP, e - EGP, ? - incomplete
Network From Reuse Path
*d :01: i
*d / :01: i
*d :01: i
*d :02: i
*d :02: i
*d :01: i
*d :01: i
*d :01: i
*d :02: i
*d :02: i
*d / :02: i
*d :02: i
*d / :02: i
*d / :02: i
*d / :02: i
*d / :02: i
*d / :02: i
*d / :02: i
*d / :02: i

11. * / i
* i
* i
* i
* i
* i
*> i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i

12. route-server>term len 512
route-server>sh ip bgp
BGP table version is , local router ID is
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,
S Stale
Origin codes: i - IGP, e - EGP, ? - incomplete
Network Next Hop Metric LocPrf Weight Path
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
* i
*> i
* i
* i
* i
* i
* i
* i

13. Two More Interconnection Techniques in Addition to BGP
1. Aggregation
2. Using Static and default routes
Aggregation - routing domain is represented by a single IP prefix and router sits at the aggregation point.
( can have more than one router and aggregation point ).Only the aggregate is advertised out on backbone.
Example aggregation network “ Interop Net 1997 ”.
Class A /8 used in convention center. One aisle was one “Routing Information Base” RIB.
Each RIB had a common /16 prefix for example /16 was “RIB 77. ” Subnet mask /22 used within each RIB
So only a portion of each RIB was used. R
FDDI
Back Bone
RP1
RP2
/22 R R R
/22
/22 R R
/22
/22
/22

14. RIPV1 was used on the RIB because all vendors routers implement RIPv1.
OSPF was used on FDDI backbone.
Routers RP1 and RP2 spoke both OSPF and RIPv1
RIB’s OSPF area ID was and each RIB was a stub area.
If an exhibitor wanted to run OSPF they could do so.
Routers RP1 and RP2 advertised default route onto RIB so RIB attached routers had a way out to internet.
Unnecessary to advertise any other routers onto RIB.
All routers on RIB learn RIP routes from each other as will as default routers from backbone routers.
Backbone routers RP1 and RP2 were configured (restricted) so they could only learn RIP Routes
from the /22s within /16

15. Examples of valid /22 prefixes that are members of this “RIB 77”:
/22
/22 …
/22
/22
Back bone routers will know about specific routes inside /16 from RIP.
Any /22 routes not within /16 will not be learned by this RIB.
For example /22 should be learned only by routers on RIB 101 ( /16 ).
With aggregation backbone routers RP1 and RP2 do not put in their routing tables every active
/22 prefix from the other RIBs
Backbone routers only advertised single aggregated prefix /16 on backbone.
Thus RP1 and RP2 are know to be where other backbone routers should send traffic for ..
When a packet arrives at RP1 or RP2 they look at RIP learned routes from their attached RIB
to decide how to forward packets.

16. Example Forwarding Table:
Known Prefixes Next-Hop Gateway Metric Source
/ OSPF
/ (connected) direct
/ OSPF
/ OSPF
/ OSPF
/ OSPF
/ ( connected ) direct
/ RIP
/ RIP
/ RIP
/ RIP
/ RIP

17. Which router contains the previous routing table?
Since (connected ) must be RP1 ( Primary Router for RIB 77 )
- Note that several / 16s were learned via OSPF via /22 interface which is the FDDI backbone.
These are other RIB aggregates. Their internal structure is invisible to us.
(each RIB is an OSPF area )
Note 5 routes learned from RIP and match the diagram drawn of the network.
- If router RP1 gets packet with destination it will be forwarded to