- IP Switching and Routing Essentials - Chapter 6 Path Vector Routing and BGP - PowerPoint PPT Presentation

Loading...

PPT – - IP Switching and Routing Essentials - Chapter 6 Path Vector Routing and BGP PowerPoint presentation | free to view - id: 1b021d-ZDc1Z



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

- IP Switching and Routing Essentials - Chapter 6 Path Vector Routing and BGP

Description:

Big Company's network has connections to both Giant Company and ... Banyan VINES address. 14. Network Address Family. AFI. Addressing for MPLS label information ... – PowerPoint PPT presentation

Number of Views:110
Avg rating:3.0/5.0
Slides: 51
Provided by: Saki8
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: - IP Switching and Routing Essentials - Chapter 6 Path Vector Routing and BGP


1
- IP Switching and Routing Essentials -Chapter
6Path Vector Routing and BGP
  • Ryo Sakiyama

2
IGP and EGP
  • Interior gateway protocol (IGP)
  • IGPs are protocols which operate within an AS.
  • RIP and OSPF are IGPs.
  • Exterior gateway protocol (EGP)
  • EGPs are protocols which route between ASs.
  • EGP used by the Internet is Border Gateway
    Protocol (BGP) version 4 now.

3
Purpose of Routing (1/2)
  • Example
  • Big Companys network has connections to both
    Giant Company and one of Giants suppliers.
  • Big can not want to supply service to Giant and
    Giants supplier.

Other AS 2
Other AS 3
Other AS 1
Giant has to use long route
Its nearest route.
Big Company AS
Giant Company AS
Giant Companys supplier AS
4
Purpose of Routing (2/2)
  • Purpose of routing
  • It is not simple like best way figure out.
  • The network now needs the best way without
    passing through a restricted area.
  • Restricted areas define policies, and policies
    make routing between Ass especially complex.

5
Implementing Policy
  • Policies are limitations on communication between
    systems.
  • Network can implement policies in at least three
    different ways.
  • Based on bandwidth
  • In the forwarding process
  • Policy based routing

6
Based on bandwidth
  • If Big Company wants to make sure that Giant
    Company uses no more than 56 Kbps of bandwidth in
    its network

Other AS 2
Other AS 3
Other AS 1
56 Kbps link limits traffic between ASs
Big Company AS
Giant Company AS
Giant Companys supplier AS
  • Giant must conform to Bigs policy.
  • Big can simply decline to connect with Giant.

7
In the forwarding process
  • Routers can implement policies by constraining
    their forwarding decisions.

Other AS 2
Other AS 3
Other AS 1
Big Company AS
Giant Companys supplier AS
Giant Company AS
discard outgoing packet if source is Giant
Company
discard outgoing packet if source is not
Big Company
  • Routers must check every signal datagram.
  • Management of packet filter requires in-depth
    knowledge of message and application protocol.

8
Policy-based routing
  • Policy-based routing places constraints on how
    routers distribute routing information.
  • If Giants routers do not know that the supplier
    lies on the other side of Bigs network

Other AS 2
Other AS 3
Giant Companys supplier AS
Giant Company AS
Other AS 1
Big Company AS
Big does not advertise any routes to Giants
supplier, even though a path is available
  • Giants routers find an alternate path to reach
    the supplier, and they forward traffic along that
    path.

9
Influencing Routing Information (1/2)
  • BGP the is protocol that conveys routing
    information based policy.
  • Routers exchange routing information with each
    other using BGP.

Routing information from top router
Top router
Routing information from left router
Right router
Routing information from bottom router
left router
The right router stores routing information from
each of its neighbors in its incoming database.
Bottom router
10
Influencing Routing Information (2/2)
  • Router create a database of local routing
    information based policy from collected routing
    information.
  • Router decides what information BGP should send
    to its neighbors.

Info to send to top router
Info from top router
Info to send to left router
Policy
Policy
Info from left router
Policy
Policy
Info to send to bottom router
Policy
Policy
Info from bottom router
11
Path Vector Routing
  • Routing between ASs look like routing with in AS.
  • Why another protocol required?
  • Both OSPF and RIP attach a metric or cost to
    possible path.
  • Exterior routers figure out how to get from here
    to there, subject to certain restrictions.
  • Routers using OSPF or RIP cannot calculate paths
    consistently, and routing loops are almost
    certain to area.

12
Distance Vector Routing without Distance (1/2)
  • BGP solve this problem by eliminating metric.
  • BGP routing algorithm is known as path vector
    routing
  • Its approach is essentially that of a distance
    vector protocol.
  • But without explicit distance or cost.
  • The essential difference between distance vector
    and path vector routing is the information
    exchanged in routing updates.

13
Distance Vector Routing without Distance (2/2)
AS 1
AS 4
AS 3
Top Token Ring
Top Ethernet
ATM Network
Frame Relay Network
AS 2
Point-to-Point Link
Bottom Ethernet
Bottom Token Ring
AS 5
14
Distance Vector Routing
AS 1
Neighbors of center router will add the cost
and select best route to destination.
AS 4
AS 3
Top Token Ring
ATM Network
Frame Relay Network
Top Ethernet
AS 2
Point-to-Point Link
Destination Distance Top Token
Ring 2 hops Frame Relay Network 1 hop ATM
Network 1 hop Top Ethernet 2
hop Point-to-Point Link 1 hop Bottom Ethernet
2 hop Bottom Token Ring 3 hop
Bottom Ethernet
Bottom Token Ring
AS 5
15
Path Vector Routing (BGP)
AS 1
Neighbors of center router will compute own
paths to destination.
AS 4
AS 3
Top Token Ring
ATM Network
Frame Relay Network
Top Ethernet
AS 2
Point-to-Point Link
Destination Path Top Token Ring
AS3, AS2, AS1 Frame Relay Network AS3,
AS2 ATM Network AS3 Top Ethernet
AS3, AS4 Point-to-Point Link AS3 Bottom Ethernet
AS3, AS5 Bottom Token Ring AS3, AS5
Bottom Ethernet
Bottom Token Ring
AS 5
16
Count to Infinity
  • Path vector provide a convenient solution to the
    counting infinity.
  • Distance vector algorithm
  • Sometimes it must count infinity to break routing
    loop.
  • Path vector algorithm
  • If routers find themselves in the path, then they
    can ignore the route.

17
Count to Infinity
  • Left router doesnt send Top Token Rings path to
    upper route, because the route has its AS number.

Destination Path Top Token Ring
AS3, AS2, AS1 Frame Relay Network AS3,
AS2 ATM Network AS3 Top Ethernet
AS3, AS4 Point-to-Point Link AS3 Bottom Ethernet
AS3, AS5 Bottom Token Ring AS3, AS5
AS 1
Top Token Ring
Frame Relay Network
AS 3
Center Router
Left Router
AS 2
18
Route Aggregation
  • There are many networks in the Internet.
  • The number of destination is large.
  • It is not practical for routers to actually track
    every separate destination in their routing
    tables.
  • BGP supports route aggregation.
  • BGP router combine multiple destination and
    create a single advertisement for all of them.
  • It also shrinks the network overhead that route
    update packets require.

19
Route Aggregation
Prefix Bits Prefix Value Binary Representation
17 172.16.0.0/17 10101100 00010000 0xxxxxxx xxxxxxxx
17 172.16.128.0/17 10101100 00010000 1xxxxxxx xxxxxxxx
16 172.16.0.0/16 10101100 00010000 xxxxxxxx xxxxxxxx
AS 1
Destination Path 172.16.0.0/17
SequenceAS1
172.16.0.0/17
AS 3
Network
172.16.128.0/17
Destination Path 172.16.0.0/16
SequenceAS3, SetAS1,AS2
Destination Path 172.16.0.0/17
SequenceAS2
AS 2
20
Normal BGP Operation (1/2)
  • BGP routers communicate with each other over TCP
    connection.
  • Once two BGP routers establish a TCP connection,
    they introduce themselves and then exchange their
    complete routing tables.
  • An OPEN message carries the introduction.
  • UPDATE messages transfer routing tables.

21
Normal BGP Operation (1/2)
  • NOTIFICATION messages that routers can send each
    other to include error information.
  • KEEPALIVE messages for routers to transmit
    periodically when there is no other traffic on
    the connection.
  • The KEEPALIVE message is important because it
    lets routers assure each other that theyre still
    active.
  • ROUTE-REFRESH message that routers use to request
    a new copy of their peers full routing table.

22
The MD5 Signature for TCP
  • BGP implementations have even introduced a MD5
    signature option.
  • MD5 signatures are not negotiated by the peers
    during establishment.
  • The MD5 signature option protect against a
    malicious party injecting false information into
    a network of BGP routers.

23
BGP within an Autonomous System
  • BGP has several properties that are useful just
    within a single AS.
  • The most important of those properties is BGPs
    ability to manage large routing databases.

BGP
AS 1
Top Router
Left Router
Right Router
BGP
BGP
24
Interior BGP (1/4)
  • One of solution is these routers use BGP within
    AS.
  • It known as Interior BGP (IBGP).
  • They can communicate directly without a lot of
    extra traffic.
  • One of problem of IBGP
  • It can not catch routing loop because all the
    routers have same AS number in the path.

BGP
AS 1
Top Router
IBGP
IBGP
Left Router
IBGP
Right Router
BGP
BGP
25
Interior BGP (2/4)
  • Suppose the top router advertises its paths in an
    update to the right router.

Destination Path 172.16.0.0/16
SequenceAS5,AS3,AS7
AS 1
Top Router
Destination Path 172.16.0.0/16
SequenceAS1,AS5,AS3,AS7
Left Router
Right Router
  • With IBGP, ignoring the routes is exactly the
    wrong things to do.
  • The whole point of using IBGP is to let the
    routers update each other.

26
Interior BGP (3/4)
  • The right router inappropriately re-advertises
    the top routers path to the left router.

Destination Path 172.16.0.0/16
SequenceAS5,AS3,AS7
AS 1
Destination Path 172.16.0.0/16
SequenceAS1,AS5,AS3,AS7
  • When the top router fails, both the left and
    right routers believe the other has a valid path,
    and traffic circulates endlessly between the two
    routers.

27
Interior BGP (4/4)
  • All IBGP routers must establish sessions with all
    other IBGP routers in the same AS, a
    configuration known as a full mesh.
  • AS the number of routers grows, the resources
    required to support the full mesh grow even
    faster.
  • To make interior operation more scalable, BGP has
    introduced two special extensions, route
    reflector and confederation.

28
Route Reflectors (1/3)
  • A route reflector is a BGP router that breaks the
    normal rules for BGP operation by design.
  • It re-advertises routing information within the
    same AS.

AS
Router C
Router A
Router B
  • Router C is a route reflector.
  • It accepts routing information from A and reflect
    that information to B.
  • It reflects Bs information to A.

29
Route Reflectors (2/3)
  • To avoid loops, route reflection introduce the
    concept of clusters.

AS
Router G
Router C
Router A
Router F
Router B
Full Mesh Connectivity
Router D
  • Routing information that it receives from cluster
    member is reflected to all peers.
  • Routing information from outside the cluster is
    reflected only to cluster members.

Router E
Route Reflection
30
Route Reflectors (3/3)
  • Originator ID
  • To ensure that misconfiguration or other or other
    errors dont introduce routing loops, route
    reflectors tag all information they reflect with
    an originator ID.
  • This value is unique among all BGP routers.

31
Confederations (1/2)
  • AS confederations offer a different approach or
    avoiding the full mesh requirement of IBGP.
  • An AS confederation takes an existing AS and
    subdivides it into many smaller groups.
  • Each smaller group acts as a full AS on its own.

32
Confederations (2/2)
  • Each group maintains a full mesh among its
    members.

Member AS
Member AS
Full Mesh Connectivity within Member AS
Full Mesh Connectivity Between Member AS
Member AS
AS
33
(No Transcript)
34
Value Message Type
1 OPEN message to establish a peering session
2 UPDATE message to exchange routing information
3 NOTIFICATION message to report errors
4 KEEPALIVE message to maintain the TCP connection
5 ROUTE-REFRESH message to request full routing information
35
vers hlen diffserv ECN payload length payload length payload length payload length payload length
fragment identifier fragment identifier fragment identifier fragment identifier 0 DF MF fragment offset fragment offset
hop limit hop limit next hdr 6 next hdr 6 header checksum header checksum header checksum header checksum header checksum
source address source address source address source address source address source address source address source address source address
destination address destination address destination address destination address destination address destination address destination address destination address destination address
source port source port source port source port destination port destination port destination port destination port destination port
sequence number sequence number sequence number sequence number sequence number sequence number sequence number sequence number sequence number
acknowledgement number acknowledgement number acknowledgement number acknowledgement number acknowledgement number acknowledgement number acknowledgement number acknowledgement number acknowledgement number
offset offset control field control field window window window window window
checksum checksum checksum checksum urgent pointer urgent pointer urgent pointer urgent pointer urgent pointer
Marker Marker Marker Marker Marker Marker Marker Marker Marker
BGP length BGP length BGP length BGP length BGP type BGP type BGP type BGP type
BGP message data BGP message data BGP message data BGP message data BGP message data BGP message data BGP message data BGP message data BGP message data
IP header
TCP header
BGP message
36
Marker Marker Marker
length type 1 version 4
senders AS number hold time hold time
BGP identifier BGP identifier BGP identifier
37
Type Option
1 Authentication information (not currently used)
2 Capabilities supported by the sender
Code Length Option
1 4 bytes Multiprotocol support
2 0 bytes Route refresh
3 varies Cooperative route filtering
64 varies GGraceful restart
38
AFI Network Address Family
1 IP version 4 address
2 IP version 6 address
3 ISO network address (NSAPs)
11 Novell IPX address
12 AppleTalk address
13 DECnet phase IV address
14 Banyan VINES address
SAFI Meaning
1 Addressing for unicast forwarding
2 Addressing for multicast forwarding
3 Addressing for both unicast and multicast forwarding
4 Addressing for MPLS label information
39
Marker Marker Marker Marker
length length type 2 withdrawn
length
routes no longer available routes no longer available routes no longer available routes no longer available
path attributes length path attributes length
path attributes path attributes path attributes path attributes
network layer reachability information network layer reachability information network layer reachability information network layer reachability information
40
Marker Marker Marker Marker
length 26 length 26 type 2 withdrawn
length 3 prefix len 16 prefix 172.16 prefix 172.16
path attributes length 0 path attributes length 0
41
Marker Marker Marker Marker
length length type 3 error code
error subcode
error data error data error data error data
Code Subcode Meaning
1 1 Connection is not synchronized properly.
1 2 Received message had an invalid length.
1 3 Received message had an unrecognized type.
42
Code Subcode Meaning
1 1 Connection is not synchronized properly.
1 2 Received message had an invalid length.
1 3 Received message had an unrecognized type.
2 1 OPEN message had an unsupported vesion number.
2 2 OPEN message had a bad AS.
2 3 OPEN message had a bad BGP identifier
2 4 OPEN message had an supported opinion.
2 5 OPEN message authentication failed.
2 6 OPEN message had an unacceptable hold time.
2 7 OPEN message included unsupported capability
43
Code Subcode Meaning
3 1 UPDATE message had a malformed attribute list.
3 2 UPDATE message had an unrecognized attribute.
3 3 UPDATE message had a missing attribute.
3 4 UPDATE message attribute flags in error.
3 5 UPDATE message attribute length in error.
3 6 UPDATE message had an invalid ORIGIN attribute.
3 8 UPDATE message had an invalid NEXT HOP attribute.
3 9 UPDATE message had optional attribute error.
3 10 UPDATE message had an invalid network field.
3 11 UPDATE message had a malformed PATH attibute.
4 The hold timer has expired.
5 An unexpected event occurred.
6 Connection will be closed.
44
Marker Marker Marker
length type 4
45
Marker Marker Marker Marker
length length type 4 AFI 1
reserved 0 SAFI 1
46
one-byte length if 4th flag is 0
flag code length
attribute data attribute data attribute data attribute data
two-byte length if 4th flag is 1
flag code length
attribute data attribute data attribute data
47
Bit(s) Meaning
7 If 1, understanding the attribute is optimal.
6 If 1, the optimal attribute is transitive and should be redistributed even if it is not understood.
5 If 1, the attribute contains only partial information.
4 If 1, the attribute length field is two bytes size.
3-0 Reserved must be transmitted as zero and ignored on reception.
48
Code Attribute Mandatory Transitive
1 ORIGIN ?
2 AS_PATH ?
3 NEXT_HOP ?
4 MULTI_EXIT_DISC ?
5 LOCAL_PREF IBGP
6 ATOMIC_AGGREGATE ?
7 AGGREGATOR ?
8 COMMUNITY ?
9 ORIGINATOR_ID ?
10 CLUSTER_LIST ?
14 MP_REACH_NLRI ?
15 MP_UNREACH_NLRI ?
16 EXTENDED_COMMUNITIES ?
49
AS B
10.0.0.0/8
10.0.0.0/8 MED10 172.16.0.0/16 MED20
10.0.0.0/8 MED20 172.16.0.0/16 MED10
AS A
50
Value Name Meaning
FFFFFF0116 NO_EXPORT Routes should not be advertised outside of the confederation (or outside of the AS if confederations are not being used).
FFFFFF0216 NO_ADVERTISE Routes should not be advertised at all.
FFFFFF0316 NO_EXPORT_SUBCONFIED Routed should not be advertised outside of the local member AS within confederation.
About PowerShow.com