Unicast Routing

1
Unicast Routing Protocols

• Acute Communication Corp.
• Edward Jin-Ru Chen
• jzchen_at_acutecomm.com.tw

2
Content

• Introduction
• Algorithms
• Distance Vector, Link State and Hybrid
• Routing Protocols
• RIP, OSPF, IGRP, EIGRP and BGP
• Testing Issue

3
Roles in the Protocol Stack

• OSI seven layers
• Physical, Data link, Network, Transport, Session,
  Presentation and Application
• Network layer
• Provides upper layers with independence from the
  data transmission and switching technologies used
  to connect systems
• Performs switching and routing function
• Such as Internet Protocol (IP)

4
IP Version 4 Header Format
5
IP Address

• IPv4 defines a 32-bit address space
• Each host contains unique address
• Divide into network space and host space

6
IP Subnet

• Divide the subnet into class A, B or C
• Using subnet mask to represent the network
  portion of the IP address
• Can use classless subnet (only depend on the
  subnet mask)

7
Autonomous System (AS)
8
AS (II)

• Each AS has unique 16-bit number
• Inside the autonomous system using the identical
  interior or intra-domain routing protocol
• Such as RIP, OSPF
• Outside the autonomous system, using the exterior
  or inter-domain routing protocol
• Such as BGP

9
What Is Routing?
6
2
3
1
5
4
10
Why Routing Protocol?

• Static Route VS Dynamic Route
• Find the way automatically

11
Default Route

• Static route for unknown destination
• 0.0.0.0 at RIP

12
Switch VS Protocol

• Cooperation between protocol engine and
  forwarding engine
• Protocol engine
• Used to be real-time OS working over processor
• Forwarding engine
• ASIC
• FPGA
• ASIC with embedded micro-processor

13
Switch VS Protocol (II)

• Protocol engine decide the forwarding port and
  tell the forwarding engine
• Forwarding engine just follow the known
  information to forwarding the multicast packet
• Address search
• Best match (or longest match)
• Best match cooperate with cache (exactly match)
• Exactly match

14
Routing Algorithms

• Distance Vector
• Link State
• Combination of upper two classes

15
Distance Vector

• Provide the route sign
• All propagated routing information are processed
  after collected
• Keep and use the processed information
• Such as Routing Information Protocol

16
Example of Distance Vector
A
B
C
D
E
17
Link State

• Provide the road map
• All propagated information are bare information
  about the link status
• Keep the bare information and use the processed
  information
• Such as Open Shortest Path First

18
Example of Link State
A
A
5
5
3
3
1
1
B
C
B
C
5
1
5
1
2
2
D
E
D
E
19
Routing Information Protocol

• Derived from XNS
• Novell IPX also uses RIP
• Distance Vector
• uses hop count as metric
• Router broadcasts table every 30 sec
• Maximum network diameter is 15 hops
• Does not support variable-length subnet masks
• subnet mask is not contained in routing updates
• Suitable for small networks

20
RIP Version 2

• Enhancement to RIP Version 1
• RIP- 2 Messages now carry
• route tag specifies origin of route information
• subnet mask
• authentication
• next hop
• RIP- 2 can use IP Multicast to send updates
• option to use 224.0.0. 9

21
RIP Characteristics

• Bellman-Ford (Distance Vector) Algorithm
• D(i, i) 0, for all i
• D(i, j) min d(i, k) D(k, j), otherwise
• Constrained by 16 hop counts
• Periodic exchange routing information
• Routing information is similar to forwarding
  information

22
Count to Infinity in RIP
Time
23
Improving the Robustness

• Split horizon
• Simple split horizon
• Split horizon with poison reverse
• Triggered update

24
Split Horizon

• Simple split horizon
• Instability may caused by neighbors engaged in a
  pattern of mutual deception
• It is never useful to claim reachability for a
  destination network to the neighbor(s) from which
  the route was learned
• Omit routes learned from one neighbor in updates
  sent to that neighbor

25
Simple Split Horizon
Net A, Metric 3
Router
No Net A entry
26
Split Horizon

• Split horizon with poisoned reverse
• Advertising reverse routes with a metric of
  infinite (16)
• If two routers have routes pointing at each
  other, poison reverse will break the loop
  immediately.
• Disadvantage is to increases the size of the
  routing messages

27
Split Horizon with Poison Reverse
Net A, Metric 3
Router
Net A, Metric 16 (Infinite)
28
Incompleteness of Split Horizon
A
C
B
29
Triggered Update

• Whenever a router changes the metric for a route,
  it is required to send update messages almost
  immediately
• combines with the rules for computing new metrics
• The receiving router believes the new
  information, whether the new metric is higher or
  lower than the old one.

30
Triggered Update (II)
Transmit immediately
Net A, Metric n1
Router
Net A, Metric n1
Net A, Metric n
Net A, Metric n1
31
RIP Timers

• Period update timer (30 sec)
• Timeout timer (180 sec)
• Garbage-collection timer (120 sec)

32
Open Shortest Path First (OSPF)

• Link-state protocol
• Shortest path first protocol
• Distributed-database protocol
• Depending on the link information to construct
  shortest path to each destination
• Use the area to reduce link-state database size
• Use the Designated Router to reduce routing
  information traffic

33
OSPF (II)

• Equal cost multi-path support
• TOS-based routing support
• Separate SPF for each TOS value
• IP subnetting support
• Attach an IP address mask to each advertised route

34
OSPF Operation
AS sample
35
OSPF Operation (II)

• RT12 advertisement
• RT12 ? N9 1
• RT12 ? N10 2
• RT12 ? H1 10
• N9 advertisement
• N9 ? RT9 0
• N9 ? RT11 0
• N9 ? RT12 0

36
OSPF Operation (III)
From
Directed Graph
To
37
OSPF Operation (IV)
The SPF tree for Router RT6
38
OSPF Area
39
OSPF Area (II)

• Divide Autonomous System into two levels
• Area 0 (Backbone area)
• Other areas transmit summarized information into
  backbone area
• Link information stored in each router, which
  belongs to the same area, is identical

40
OSPF Area Operation
AS with Area sample
41
OSPF Area Operation (II)
From
Area 1's Database
To
42
OSPF Area Operation (III)
From
Backbone's Database
To
43
OSPF Hello Protocol

• Periodic send the hello packet containing the
  discovered neighbors
• Discover OSPF neighbors
• May use multicast (AllSPFRouters) on broadcast or
  point- to- point links or configuration may be
  required
• Elect the Designated router
• DR only elected on broadcast and point- to- point
  links

44
OSPF Hello Protocol (II)

• Establish Adjacencies between Neighboring Routers
• Only adjacent routers exchange routing table
  updates
• Use to control the distribution of routing
  information

45
OSPF Designated Router

• Elected through Hello Protocol
• High priority first
• High router ID first
• Originate network links advertisement on behalf
  of the network
• Adjacent to all other routers on the network

46
OSPF LSA

• Router links advertisements
• Network links advertisements
• Summary link advertisements
• Advertise routes to networks
• Advertise routes to AS boundary routers
• AS external link advertisements
• Type 1 external metric equivalent to the link
  state metric
• Type 2 external metric greater than any internal
  metric

47
OSPF Routing

• Synchronize Link-State Databases
• adjacent routers exchange database description
  packets
• link- state request/ updates provide neighbors
  with most recent LSA - flooded within area
• Calculate the routing table

48
OSPF Extension

• Functions and Services provided by OSPF can be
  easily extended
• define new information and use LSA advertisements
  to flood throughout routing domain
• OSPF Opaque LSA (RFC2370) designed to carry new
  information
• routers may use this information or other
  applications may use OSPF to flood data

49
OSPF Extension (II)

• Two New OSPF Services make use of the Opaque LSA
• Address Resolutions Advertisements (ARA)
• Optimized Multipath (OMP)

50
OSPF ARA

• Utilize fast and reliable OSPF topology updates
  to propagate link- layer information (IP/ ATM
  address mappings) to OSPF ATM- attached routers
• Not subject to packet loss like NHRP and no need
  to query address resolution server
• Supports point- to- point, point- to- multipoint
  and multipoint- to- point connections

51
OSPF ARA (II)

• Interoperate with existing mechanisms (MPOA,
  NHRP)
• Associate group of routers into a single logical
  network (VPN)
• attached logical network ID value to ARA packets

52
Multipath Forwarding

• More than one path of equal cost may exist
  between two points in the network (termed Equal
  Cost Multipath)
• Routing protocols such as OSPF may support this
• Multipath forwarding means that the router
  maintains multiple next hop entries for a
  destination

53
Multipath Forwarding (II)

• Forwarding can be done on a per-packet
  round-robin basis
• however different paths may exhibit different
  delay, bandwidth and MTU characteristics
• problematic for TCP sender and receiver
• possible to generate out-of-order transmission
• error loss retransmission may happen

54
Multipath Forwarding (III)

• Another technique is to divide traffic equally
  across multiple paths by applying next-hop
  identifier (hash) to each source/destination
  address pair
• Still no knowledge of load or capacity of equal
  cost paths

55
OSPF OMP (Optimized Multipath)

• Use OSPF Opaque LSA to distribute loading
  information for equal cost paths
• LSA_ OMP_ LINK_ LOAD measures load, capacity
  and packets dropped from a particular link
• LSA_ OMP_ PATH_ LOAD
• Adjust distribution of traffic across multiple
  paths based on advertised OMP loading information

56
OSPF OMP Forwarding

• Hash boundary (meaning percentage of traffic
  flowing over equal cost paths) may move depending
  on load information conveyed by OMP updates

57
Interior Gateway Routing Protocol

• Proposed by Cisco Systems, Inc.
• Is a distance vector interior-gateway protocol
• Use a combination of metrics
• Internetwork delay, bandwidth, reliability, and
  traffic load
• Reliability and load can be ranged from 1 to 255
• Bandwidth can be ranged from 1.2kbps to 10gbps
• Delay can be ranged from 1 to 2 to 24th power

58
IGRP (II)

• Permit multipath routing
• Dual equal-bandwidth lines may run in round-robin
  fashion, with automatic switch over to other when
  one line goes down
• Multipath can be used even with different metrics
  (if bandwidth is 31, offered load set to be 31)

59
IGRP Stability Features

• Hold-downs tell routers to hold down any changes
  that might affect routs for some period of time
  to avoid the update information polluted by
  regular update
• Split Horizon
• Poison Reverse Updates

60
IGRP Timers

• Update timer
• The time to send the routing update message (90s)
• Invalid timer
• The time to decide the route invalid without
  refreshed information
• Hold-time period
• Flush timer
• Time to flushed from the routing table

61
Enhanced IGRP

• Combination of link state protocol and distance
  vector protocol
• Using Diffusing Update Algorithm (DUAL)
• Fast convergence
• Store all of its neighbors routing table
• If not appropriate route exists,queries its
  neighbor for an alternate routes
• Variable length subnet mask

62
Enhanced IGRP (II)

• Parital, bounded updates
• No periodic update
• Send partial updates only when the metric for a
  route changes (Less bandwidth requirement)
• Multiple network-layer support
• AppleTalk, IP, and Novell NetWare
• Redistribute routes learned from OSPF, RIP,
  IS-IS, EGP, or BGP. Novell implementation
  redistributes routes learned from Novell RIP or
  SAP

63
Enhanced IGRP (III)

• Features four new technologies
• Neighbor discovery/recovery
• Using hello packet
• Reliable Transport Protocol
• For update and acknowledgement not for hello
  packet
• DUAL finite state machine
• Protocol-dependent modules

64
Inter-Domain Routing

• Policy Routing - Deciding where to direct
  information based on
• Cost, Performance, Security, Availability and
  Reliability, Traffic Type - Best Effort or Real-
  time, Others...

65
Border Gateway Protocol

• Designed as a true inter- AS routing protocol for
  TCP/ IP- based networks
• Uses concept of Path Vectors to represent path to
  reachable destination
• prevents loops
• Enables policy- based routing by affecting route
  selection and controlling the distribution of
  specific routes

66
BGP (II)

• Uses TCP to reliably exchange routing information
• BGP4 supports route aggregation and variable
  length subnet masking
• Inter- BGP Router relationships
• Internal BGP between two BGP routers within
  same AS
• External BGP between two BGP routers in
  separate AS
• No restrictions on network topology
• RFC1771

67
BGP Path Vector
AS 2
Net A, Path 1
Net A, Path 1,2
AS 1
AS 3
Net A
Not accept
Net A, Path 1,2,3

• BGP routers advertise routing information which
  contains a sequence AS numbers that a route has
  traversed. This is referred to as a Path Vector
• A BGP router will not accept an update if it sees
  its own AS number in the update
• This ensures loop free inter- domain routing

68
BGP Routing Process
69
BGP Routing Process

• Routing updates are received from other BGP
  routers
• Input policy engine filters routes and performs
  attribute manipulation
• Decision process decides what routes BGP router
  will use
• Output policy engine filters routes and performs
  attribute manipulation for routes to be
  advertised
• Routing updates are advertised to other BGP
  routers

70
BGP Message Flow

• BGP peers establish a TCP connection with each
  other
• Initially the entire routing table is exchanged
  after that only changes in topology or policy are
  sent in UPDATE messages
• BGP Updates can announce or withdraw a route
• BGP Updates also carry attributes which are used
  by the policy engines and the decision process
• AS_ PATH, ORIGIN, NEXT_ HOP, MULTI_ EXIT_ DISC,
  LOCAL_ PREF, etc.

71
Protocol Verification

• Packet Format
• Lower protocol parameter setting
• Entry field validity
• Timer
• Preciseness of each timer
• Algorithm
• Using entered packet to generate virtual
  environment to trigger algorithm calculation

72
Protocol Verification

• Input process
• Check the processing result of different input
  packets
• Output process
• Check the processing result when router generate
  packets

73
Testing Example

• RIP timer verify
• Divide the RIP process into slots
• Slots is separated by the periodic update
• Procedure
• Transmit a response packet into DUT (Device Under
  Test)
• Count the number of periodic updates contains the
  newly added entry
• Verify the time to become invalid and disappear

74
Timer Verify
180 sec
120 sec
180 sec
120 sec
Periodic Update
Response
Triggered Update
75
Benchmarks

• Throughput (pps)
• Routing entry update delay