Open Shortest Path First Fulvio Risso Politecnico di Torino This set of slides is based on a previous version created by Mario Baldi and Giorgio Valent

1

General concepts 

Link State protocol defined by the IETF 

OSPF v.1: RFC 1247 (1991)



OSPF v. 2: RFC 2328 (1998)



Newest versions exist for IPv6 (OSPF v.3: RFC 5340, 2008)



OSPF has the hierarchical concept:

3



Can handle large networks



An AS is subdivided in areas



Each area contains a group of contiguous networks



Backbone: special area, not necessarily contiguous, which is connected to all other areas

Example of a possible OSPF network Autonomous System boundary router (ASBR)

R1

Internet Area 1

R2

R3 R4 R5

Internal router

R6

Area 0 (backbone)

R9

R7

R10

Area 2

R11

R12

Area border router (ABR)

R8

R13

Area 3 R14

4

Backbone router

R15

Terminology (1) 

OSPF defines its own terminology, which is not always aligned to what other protocols do 

Sometimes names are misleading, e.g. “autonomous system boundary router” is not necessarily the router at the border of an AS



Backbone  level-2 network



Backbone router  router in a level-2 network



Area border router  router that has interfaces in two or more areas (one of them must be the backbone) 

This router executes many copies of the link state algorithm (one per area)



Internal router  router in a level-1 network



Note: level-1 and level-2 refer to the hierarchical placement of those network, not to the OSI layer

5

Terminology (2) 

Autonomous System (AS)  OSPF routing domain



AS boundary router  router that has at least one interface connected to another routing domain (not necessarily into another AS)



Link State Advertisement (LSA)  data structure that contains the “core” link state information

6



LSA are not packets



A router can put several LSAs within the same packet

Metrics 

OSPF supports more than one metrics simultaneously on a single link



The best path may be, depending on the packets



7



The shortest path



The one with the best capacity



The one with the lowest delay



…

OSPF allows to define metrics depending on the TOS field of the IP packet 

In theory, 64 possible types of service



In practice, this feature is almost unused

Costs 

Differently from IGRP, OSPF does not define an unambiguous way to calculate the cost of a link



Assigned by the manufacturer of the network device 

Between 1 and 65535



Each manufacture has his own default values Cost (Cisco) = 108/bandwidth



Better to customize it on most important links (on both ends) interface ip ospf cost



Also possible to change globally for the router (defaut: 100) router ospf auto-cost reference-bandwidth

8

Equal Cost Multi-Path Routing 



Possible when more than one path have the same cost 

Also known as Load Sharing



Note that paths that have the same cost may not be, in fact, equivalent from the point of view of the network provider

OSPF does not mandate the implementation of ECMP Routing 



Almost all implementations support ECMP Routing 

9

OSPF does not preclude the possibility to use several paths in parallel if they have the same cost but it does not give further details

Each implementation may have different criteria for distributing the traffic across the equivalent links

OSPF and aggregation 



The routers of the backbone are usually configured (manually!) in order to aggregate network addresses 

Network summaries are propagated in the other areas



For example

10

An area contains addresses 5.12.*.* and 5.7.*.*



The level 2 router propagates the connectivity toward 5.*.*.*

Backbone routers propagate summaries of the networks received from the other areas 





Routers inside an “edge” area may not choose an optimal exiting point from the area

The aggregation must be specified manually by the operator in order not to have troubles with network summarization

Router ID 

Unique “name” of the router



OSPF does not specify how it should be determined 



It mandates that the algorithm produces unique identifiers

Cisco 

Bigger addresses present on the loopback interface 



If there is no loopback interface, the bigger IP address configured on the OSPF interfaces is chosen



The RouterID is computed at the beginning of the OSPF process and is not modified even if the IP addresses on the router are modified 

11

Loopback interfaces do not depend on the state of the physical interfaces and are thus more stable

In such a case, the router may have a new RouterID at the following reboot

Authentication 

OSPF can store information for authenticating the other end party



To each link is associated one (and only one) password 

Only the OSPF packets containing the password associated to the link they come from are accepted



It is difficult to change the password of a link 

12

You have to change on both ends of the link



Easy to intercept a message to have the password of a link and to use it in both directions



Almost unused, due to its weakness

Propagating LSA on the network 

Routers have to propagate LSA on the network 



Needed e.g., to propagate topological changes to the rest of the network

In practice, LSAs are generated: 

When a change is detected in the network (e.g., cost, link up/down, etc.)



When a timer expires (default: 30 min) 





13

This makes the network much more stable, even if LSAs are acknowledged It prevents an LSA of a dead router to keep staying in the network forever

Please note that LSAs in the OSPF database expire if not renewed within a given time (default: 60 min)

Selective flooding protocol 

Selective Flooding protocol is used to propagate LSA



Flooding requires includes two types of packets 

Link State Update (Header OSPF: Type = 4)



Link State Acknowledgement (Header OSPF: Type = 5) 



14

Needed to implement a reliable transmission of the LSAs and guaranteeing that the database is coherent across all the routers

In practice, two fields are needed in the OSPF packet: 

Sequence Number (to recognize the newest LSA)



Age (to delete old LSA that have not been renewed)

OSPF Sequence Numbers (1) 



OSPF v.1 uses a lollipop space 

Studies demonstrated that the overflow in OSPF was not a real problem



Sending a new LSA every 1 sec. with a SeqNum over 32 bits translates into 136 years of continuous running before getting an overflow

OSPF v. 2 uses a linear space 

15

A special rule applies when the SeqNum reaches the end of the numbering space

OSPF Sequence Numbers (2) 

Value -N (0x80000000) is reserved (and unused)



Value -N + 1 (0x80000001) represents the Initial Sequence Number



Afterwards, the LSA sequence number is incremented each time the router originates a new instance of the LSA



If a router R1 receives a LSA with an age < of the age of the LSA currently in its database, the “newest” LSA is flooded to all the network (hence it will reach also the sender R2) 

16

In case R2 is the originating router for that LSA, it recognizes that an “older” LSA exists in the network, hence it updates its SeqNum to the one of the received LSA + 1 and it re-floods it

OSPF Sequence Numbers (3) 



17

If a router has to send a LSA with value N - 1 (0x7fffffff; also referred to as MaxSequenceNumber), the current instance of the LSA must first be flushed from the routing domain, which is done by prematurely aging the LSA 

The LSA is sent with Age equal to the maximum value MaxAge



Any receiving router has to flood that LSA as usual, then it keeps it in memory until all the ACKs from its neighbors have been received



At that point, that LSA is deleted from the memory of the router

As soon as this flood has been acknowledged by all adjacent neighbors, a new instance can be originated with sequence number of InitialSequenceNumber

LSA on broadcast networks (1) 

OSPF has to implement a special algorithm for propagating LSA on broadcast networks 





High complexity for Dijkstra, a lot of traffic exchanged over the link, many (useless) synchronizations when a new router joins the network

On each broadcast network, two routers are selected 

Designated Router (DR)



Backup Designated Router (BDR)

Election process based on the two routers that have the best RouterID 

In practice, the two routers that boot first become DR and BDR 

18

The election process is initiated as soon as two routers are found on that network, and DR/BDR do not decade unless an explicit configuration command is given

LSA on broadcast networks (2) 

When a router R has to transmit a LSA on the LAN, it sends the packet to the multicast data-link address AllDrouters 

Only the DR and the BDR join that group, hence they receive traffic at the AllDrouters multicast address



The ACK is generated by the DR and sent back to the original sender at the AllSPFrouters multicast address



Now, the DR forwards the LSA to the multicast data-link address AllSPFrouters



The routers send the acknowledgment reception) to the address AllDrouters 



19

(that

confirms

This is needed in order to allow the BDR to receive the ACK

If the DR does not receive acknowledgment from a subset of routers, it sends copies of the LSA to each of these routers (in unicast)

Propagation of the LSA on LAN 

The DR keeps a lot of information about the state of the other routers



The replacement of the DR would need:





A huge exchange of messages needed by the new DR to gather this information



A “large” amount of time before being actually operative

To avoid so, a Backup Designated Router (BDR) is elected 

Sends messages directly to the DR



Records the same information gathered by the DR



DR and backup DR are never replaced by another router, unless they stop working



Beware: the function of DR/BDR is determined per-LAN

20



A router can be DR in a LAN and nothing in another



This is a property of the router interface, based on the RouterID and the Priority

Partitioned areas (1) 

Classical problem of the hierarchical routing



In OSPF, this is handled differently depending on the type of area 

A normal area



The backbone

A

B

C

Area 1

F

G Area 2

Backbone

D

21

L

E

H

I

Partitioned areas: Area (2) 

Once the routes are summarized, a packet from I to B may be injected in the Area 1 from C or from E 

The best path is I-H-E-D-B



If it is injected from E, the packet cannot reach its destination

A B

C

L

Backbone

F

Area 1 D

22

G Area 2

E

H

I

Partitioned areas: Area (3) 

Solution: the Area Border Router does not summarize the information about all the networks present in the edge area; instead, it announces only those network that are reachable from the node itself 

23

In the backbone there is still only one entry for each network, in order to be able to select the best path when 2 border routers announce the same network (as it is the case for the network B when there is no failure)

Partitioned areas: Backbone (1) 

A failure in the backbone may isolate the router E even ifan alternative route through area 1 may exist



The solution is to create a Virtual Link between E and another node of the backbone (for example C, by going through D and B)

A B

C

F

Area 2

Area 1 D

24

G

E

Backbone

H

I

Partitioned areas: Backbone (2) 





25

The Virtual Link looks like a tunnel whose endpoints are automatically configured by the OSPF 

Configuration is automatic, but…



…activation must be done by hand

Virtual Links are a sort of automatic tunnels 

In fact, a GRE tunnel looks similar, but the entire configuration has to be done by hand



OSPF will advertise one more link crossing the backbone (the Virtual Link, in fact), although its cost is usually rather high

Advanced use of Virtual Links: bring into the backbone a router that is not directly connected to other backbone routers

Partitioned areas: Virtual Link 

OSPF routing messages are encapsulated in IP unicast packets crossing the link 



The destination and the source are the endpoints of the virtual link (C and E)

The data packets sent from C to E have to go through area 1 

The level 1 routers have to know that the packets are heading to E 



Note: the configuration of the VL requires only the RouterID of the two backbone routers involved, and not the IP addresses of their interfaces 

26

The information must come from the level 2 information propagated inside the area

OSPF will derive the correct IP address automatically

Timers 

Key values announced in the Hello packet 

HelloInterval: 10s



RouterDeadInterval: 40 sec 



Ignored if the router detects the death of one of its neighbors through a signal coming from the physical level

The adjacency of a router with another peer is established only if the values of the timers are the same for both



Those parameters have an impact on the convergence speed of a network



Non specified (only suggested) by the standard



Other timers

27



LSA Refresh: every 30 min



MaxAge: 1 hour

Content of the OSPF database 

The database of a router belonging to an area contains: 

Link States belonging to all the routers in this area



Summary Records, generated by the ABR and related to all the networks belonging to the other areas (including the backbone); those routes can be summarized 



Previous Links States are never summarized

External Records, generated by the ASBR related to all external routes



This splitting contributes to the scalability of OSPF



Database entries are removed if replaced by a new entry, if invalidated by special message (an entry with MaxAge) or in case of timeout

28

Link State Records: types 

In the LS-like algorithms, every adjacency is modeled as a point-to-point link 



In the case of a LAN, the number of adjacencies may explode 



Adjacency = adjacent router, adjacent network IP

A particular adjacency is defined with the Designated Router

OSPF defines two types of link 

Router Link  



Point-to-point connection between a router and an adjacent IP network

Network Link 

29

Point-to-point connection between two routers (e.g., serial link)

Point-to-point connection between a router an a transit network that hosts two or more OSPF routers

General structure of OSPF packets IP, Protocol #89

OSPF Header

OSPF packet

OSPF Data

Hello

Database Description

Link State Request

Link Type

Link ID

Link Data

1 (Point-to-point link to another router)

Neighboring router’s RouterID

IP address of the originating router’s interface to the network

2 (Link to a transit network)

IP address of the DR’s interface

IP address of the originating router’s interface to the network

3 (Link to a stub network)

IP network address

IP network mask

4 (Virtual Link)

Neighboring router’s RouterID

The MIB-II ifindex value for the originating router’s interface

Link State Update

LSA Header

Router LSA Network LSA Network Summary LSA

ASBR Summary LSA AS External LSA Link State Acknowledgement

30

Link State in OSPF (1)

NetA/MaskA

NetA/MaskA IP1

R1

Link State Database: Router LSA (Link State ID: R1, Adv Router: R1, #links: 1) Router Link: Link ID (Network ): NetA

31

Link Data (Netmask): MaskA

- Link to a stub network

Link State in OSPF (2) R1

R1

IP1

IP1

NetA/MaskA

NetA/MaskA NetA/MaskA

IP2

IP2

R2

R2

Link State Database: Router LSA (Link State ID: R1, Adv Router: R1, #links: 2) Router Link: Link ID (Network) : NetA Router Link: Link ID (Neighbor RouterID): R2

Link Data (Netmask): MaskA Link Data (Router If. Addr.): IP1

- Link to a stub network

- Point-to-point link to another router

Router LSA (Link State ID: R2, Adv Router: R2, #links: 2) Router Link: Link ID (Network): NetA Router Link: Link ID (Neighbor RouterID): R1

32

Link Data: MaskA (Netmask)

- Link to a stub network

Link Data: IP2 (Router If. Addr.) - Point-to-point link to another router

R1 IP1

Link State in OSPF (3)

R2 IP2

R1

IP1 NetA/MaskA IP2

R2

Transit Network (includes NetA/MaskA)

IP3

IP3

R3

R3

Router LSA (Link State ID: R1, Adv Router: R1, #links: 1)

Link State Database:

Router Link: Link ID (DR IP Address): IP3

Link Data (Router If. Addr.): IP1

- Link to a transit network

Router LSA (Link State ID: R2, Adv Router: R2, #links: 1) Router Link: Link ID (DR IP Address): IP3

Link Data (Router If. Addr.): IP2

- Link to a transit network

Router LSA (Link State ID: R3, Adv Router: R3, #links: 1) Router Link: Link ID (DR IP Address): IP3

Link Data (Router If. Addr.): IP3

- Link to a transit network

Network LSA (Link State ID: IP3, Adv Router: R3) Netmask: MaskA

33

Attached Routers: R1, R2, R3

OSPF: LS topology and hierarchy 

Each LS router knows perfectly the topology of the area it belongs to



With respect to other areas: 

The precise topology is unknown



The router can know the list of destinations reachable outside its area 



A router that belongs to many areas has many OSPF databases, one per each area 



34

Some destinations may be summarized (e.g., default route for stub areas)

Of course, this will originate a single routing table

Some examples of the view from a router are the next slides

View of the network from a single router (1) Net1 Net1 (Transit network) R1

R2

Area 0

Net 2

Net3

R4

R1

R2

Area 0

R3 Net2

R3

Net3

Net4

Area 1 Net5 R6

R5

Net4

Net5

Net6 Outside Area 0

Net6

Net7

Complete network topology 35

View from R2

Net7

View of the network from a single router (2) Outside Area 1

Net1 R1

R2

Net2

Area 0 Net1

Net 2

Net3

R4

Area 1

Area 1 Net5

R4

Net5 (Transit network)

R6

R5

R6

R5 Net6

Net7

Complete network topology 36

R3

R3

Net4

Net6

Net3

Net7

View from R4

Net4

Net4

View of the network from a single router (3) Outside Area 1 Net1 (Transit network) Net2 R1

R2

Net2

Area 0

Net1

R3

R3

Net3

Area 1 Net4

Net3

Net5 Net6

R4

Net5 (Transit network) Net7

Net4

Net4

R6

R5

Outside Area 0 Net6

View from R3 (in area 0) 37

Net7

View from R3 (in area 1)

OSPF packet format 

It is encapsulated directly in IP (protocol Type = 89)



All fields are fixed length



There is no room for keeping future extensions





38



Executable code (and parsing) is more compact



Packets processing is faster



More rigid to evolution

It is in fact split in 3 sub-protocols 

Hello



Exchange



Flooding

All packets begin with the same header

Packet format: header (1) 0

8

Version

16

31

Type

Packet Lenght

Router ID Area ID

Checksum

Authentication Type Authentication

39

Packets format: header (2) 

Version: nowadays 2



Type: may have the following values:





Hello, Database Description, Link State Request, Link State Update, Link State Acknowledgement



Represents the type of the transported OSPF packet

RouterID: Router ID of the device that propagates the message 

40

E.g., in case of flooded packets, the RouterID represents the router that propagates the message, not the one that actually generated the LSA

Packet format: header (3) 



41

AreaID: number that identifies uniquely the area inside the OSPF domain 

An IP address is often chosen



The value 0 identifies the backbone

Authentication Type: two possible options exist: 

No Authentication



Simple Password

Hello packets (1) 

Used to: 

Test whether the link is fully active, thus to build the table of adjacencies 





In order for a link to be active, bidirectional connectivity must be verified By-product: OSPF does not work on unidirectional (e.g., some satellite) links

Elect the Designated Router and the Backup DR on the LAN



Hello packets are only transmitted to neighboring nodes and never propagated



Header OSPF: Type = 1

42

Hello packets (2) 0

8

16

31

Network Mask Hello Interval

Options Dead Interval

Designated Router Backup Designated Router Neighbor1 ....... NeighborN 43

Priority

Hello packets (3) 

Network Mask: the netmask associated with the interface that generated the Hello



Hello Interval: time between two consecutive Hello packets



Options: Only the two last bits are defined 

E: if the router is able to send and receive external routes; it is equal to 0 of the interface is part of a stub area 





T: if the router is able to handle packets with different TOS

Priority: used for the election of the DR 

A router without the Priority bit cannot become DR 



44

If two routers have different values for this bit, the adjacency cannot be established

The best router with the Priority bit is elected DR

Set (manually) by the network manager

Hello packets (4) 

DeadInterval: maximum temporal validity of the Hello packet 



DR, BDR: address of the Designated Router – Backup DR 



45

Set to 0 if the election process has not been completed (or no need to elect any DR/BDR)

Neighbor: list of RouterID that indicate the routers that are directly reachable from that network interface 



If no Hello packets are received from a neighbor passed this time, the router is considered dead

I.e., the routers whose Hello packets have been received on that interface over the last DeadInterval seconds

The link between two routers is declared operant if 

Packets can flow in both directions



Both routers have the same value of the bit E

Exchange protocol (1) 



Requires four types of packets 

Database Description (Header OSPF: Type = 2)



Link State Request (Header OSPF: Type = 3)



Link State Update (Header OSPF: Type = 4)



Link State Acknowledgement (Header OSPF: Type = 5)

Used to synchronize the database of two routers when they become adjacent 



It is an asymmetrical protocol 

46

For example at boot time or when a new link becomes active

The first step consists in the definition of which router becomes the Master and which one acts as Slave

Exchange protocol (2) 

Phases of the Exchange protocol 

Definition of the Master/Slave roles



Exchange of a summary of all the LSA available on the routers on both sides of the connection



Request to the other router of old/missing LSA



Transmission of the requested LSA 



The normal sending of the LSA (outside the exchange protocol) looks definitely similar to how this protocol operates 

47

The complete LSA is transferred only at this step (if needed)

In this case, only two types of packets are used 

Link State Update



Link State Acknowledgement

Database Description Packets (1) 0

16

0

24

Options

0

DD Sequence Number Link State Type Link State ID Advertising Router Link State Sequence Number LS Checksum

LS Age ............

48

2930 31

I M Ms

Database Description Packets (2) 

Options: as in the Hello packet 

E: if the router is able to send and receive external routes; it is equal to 0 if the interface is part of a stub area



T: if the router is able to handle packets with different TOS



I: Initialize



M: More



MS: Master - Slave (1= Master)



DD SN: sequence number in the Database Description packet



The remaining fields (that can be repeated) are the description of the header of a LSA and thus have the same meaning already presented for the LSA

49

Exchange Protocol: initialization (1) 

The router that wants to initialize the procedure 





The other router answers 

Emits a DD packet of “acknowledgment” with I, M



If its RouterID is better than the one received, it sets the MS bit as well, and it becomes the Master



If not, it accepts the other as Master and it keeps the MS bit to 0

The first router begins sending the summaries of its LSA 



50

DD packets with M, MS, till last packets that has only MS

The slave answers 



Emits an empty DD packet with I, M, and MS set.

Emits a DD packet of “acknowledgment” with M, reporting its “own” description of the database

If the master does not receive the Ack within a given timeout, it resends the original DD packet

Exchange Protocol: initialization (2) 



51

If vice-versa the slave has not finished to transmit its descriptions 

In correspondence to the packet of the Sender with M=0, it emits a DD with M=1



The master continues to send empty packets with M = 0, and to accept the Acks that come from the slave



The procedure of synchronization ends when also the slave sends a packet with M = 0

During the exchange 

Both the master and the slave checks that they have the LSA sent by the counterpart and that this LSA is not older than the one received



If this is not verified the LSA is inserted in the list of the LSA that needs to be fully exchanged

Link State Request packets 

They are sent at the end of the DD if there are some LSA that need to be synchronized 



More requests can be grouped together 



They request the other router to send the complete LSA corresponding to the field Link_State_Type, Link_State_ID and Advertising_Router indicated The three field may be repeated more than once in an OSPF packet

Requested LSA are sent using the flooding protocol 

In this way it is possible to update the whole network with the missing information

0

31

Link State Type Link State ID Advertising Router 52

Link State Update Number of Advertisement: the number of LSA that are transported in the current packet





A single OSPF can contain multiple LSA, even of different types

0

31

Number of Advertisements N

LSA

53

LSA Header (1) 

All LSA have the same header



Advertising Router: RouterID of the router that generated the LSA 

0

Not the one that propagated the LSA

8

16

LS Age

31

Options

LS Type

Link_State_ID

Advertising Router LS Sequence Number

LS Checksum 54

Length

LSA Header (2) 

Age: age of the current LSA (in seconds)



Options 

E: External Link



T: set when the router supports the Routing TOS



Type: type of transported LSA (Router LSA, Network LSA, etc.)



LinkID: identifier whose meaning depends on the type of LSA

55



Chosen by the Advertising Router, but the exact meaning may change according to the Type



The combination of RouterID, identifies univocally a LSA

LinkStateID

and

LSA

Type

OSPF LSA types 

LSA 1 – O, Router LSA 



LSA 2 – O, Network LSA 



Generated by ASBR and is propagated between areas

LSA 7 – O N1, O N2, NSSA External LSA 

56

Generated by the ABR and is propagated between areas

LSA 5 – O E1, O E2, External LSA 



Describes a network from another area, generated by ABR and is propagated between areas

LSA 4 – O IA, ASBR Summary LSA 



Contains all routers attached to the segment, generated by DR and is local to the area

LSA 3 – O IA, Network Summary LSA 



Contains all Link IDs – network, generated by every router and is local to the area

Generated by ASBR into NSSA area and is propagated into area 0 as E1 or E2

Link State Advertisement 

It represents the real data structure that transports the Link State (as presented in the theory)



Each LSA begins with a common header



The remaining part is specific for each LSA type

57

Link State Acknowledgement 

Message used to confirm the successful reception of an LSA (through the LS Update)



Sent in unicast (unless we are on a broadcast network)



Used both in the normal flooding process and in the Exchange phase 0

16

31

Link State Type Link State ID Advertising Router Link State Sequence Number LS Checksum 58

LS Age

LSA Type 1: Router Link (1) 



It keeps the information about all the links connected to the router that is currently creating the LSA 

The information includes all the adjacent routers and all connected LANs



The LSA includes N repeated sections, one for each “link” of the router

Propagated only inside the current area 

59

No matter if the area is a backbone or an edge area

LSA Type 1: Router Link (2) 0

6 7 8

0

EB

16

0

31

Number of links

Link ID Link Data Link Type

# TOS

TOS 0 metric

TOS = x

0

TOS x metric ........

TOS = z 60

0

TOS z metric

LSA Type 1: Router Link (3) 

E: set if the router is an Area Border Router (External)



B: set if the router is an AS Boundary Router (Border)



Link Type, LinkStateID, LinkData: presented in Slide 30



A specific metric for all the TOS that are used in that network is stored

61



This will originate N Shortest Path Trees, with the obvious consequences in terms of memory occupation and CPU consumption



Only the TOS whose metric is different from the value associated to TOS 0 (which represents the default 0) is explicitly stored in the packet



TOS=0 always exist



Other TOS are assumed to be equal to the cost of TOS=0

LSA Type 2: Network Link (1) 

Type = 2



Generated by the Designated Router on behalf of the transit network 

Lists all routers present on the LAN

0

31

Network Mask Attached Router 1

....... Attached Router N

62

LSA Type 2: Network Link (3) 

Network Mask: netmask of the transit network



Attached Router: IP address of all routers that it is interfaced with in the LAN (or rather, all the routers that have created an adjacency with the given router) 



63

It is not mandatory to indicate the number of adjacent routers because the length of the message is given by the field LENGTH of the header

LinkStateID (in the LSA header): IP address of the interface of the Designated Router connected to the transit network

LSA Type 3-4: Summary Links (1) 

Generated by the ABRs and aim at providing summaries for IP networks and ASBRs



Both deal with summarized information



Each LSA includes only one destination (differently from LSA Type 1 - Router Links) 

64

Each ABR generates a lot of them, inserted in the same packet

LSA Type 3-4: Summary Links (2) 

Summary Links for IP Networks (Type = 3) 

Propagates the distance between the ABR and the IP networks present in the AS



Generated only by ABRs



An ABR generates different LSA for each area it belongs to 



65

Obviously, summaries are different according to the area we are looking at

Summary Links for Border Routers (Type = 4) 

Propagate the distance between the ABR and the ASBR



Used to determine the best ABR for destinations (i.e. networks outside the AS)



Generated only by ABRs



An ABR generates exactly the same Type-4 LSA on all the areas it belongs to

reaching

external

LSA Type 3-4: Summary Links (3) 0

8

16

31

Network Mask TOS = 0

0

TOS 0 metric

TOS = x

0

TOS x metric ........

TOS = z

66

0

TOS y metric

LSA Type 3-4: Summary Links (4) 



Network Mask 

Type 3: netmask associated to the summarized IP network



Type 4: set to 0xFFFFFFFF

TOS = x, TOS x metric: same meaning as the one of the Router Links 



67

The field # TOS is not used because the length can be found using the LSA header

LinkStateID (in the LSA header) 

Type 3: the network address of the summarized network



Type 4: IP address of the Area Border Router

LSA Type 5: External Links (1) 

Type =5



Keeps the cost to reach the external destinations, calculated from the ASBR



68



Generated by the AS Border Router



Propagated to all routers of the OSPF domain

It includes one destination for each LSA (as for the Summary Links)

LSA Type 5: External Links (2) 0

8

16

31

Network Mask E, TOS = 0

0

TOS 0 metric

External Route Tag (0) E, TOS = x

0

TOS x metric

External Route Tag (x) ........ E, TOS = z

0

TOS y metric

External Route Tag (z) 69

LSA Type 5: External Links (3) 

Network Mask: netmask of the destination network



E: bit 0 of the field TOS

70



Indicates if the external routes are learned by protocols such as BGP, which do not necessarily supplies a metric comparable to the one of OSPF



If it set, it indicates that the metric is not compatible with OSPF and thus the distance must be considered “bigger than any other internal route”



In the case in which the metric is compatible, it is possible to add it to all the costs inside the OSPF domain, hence getting the real cost to the destination

LSA Type 5: External Links (4) 

External Route Tag: used by the Border Router to exchange information related to this route 



TOS = x, TOS x metric: same meaning as the one of the Router Links 



71

It is neither used nor analyzed by OSPF

The field # TOS is not used because the length can be found using the LSA header

LinkStateID (in the destination network

LSA

header):

IP

address

of

the

OSPF: different types of areas 

Normal area 



ABRs forward all LSAs from Area 0, including external LSAs

Stub Area 

Eliminates external routers/LSA present in area (Type 5)



External routes are replaced by a single Default Route

! ! Normal area router ospf 100 ! !

72

! ! Stub area router ospf 100 area 10 stub !

OSPF: different types of areas (Cisco extensions) 





Totally Stubby Area 

Keeps only internal LSA (Type 1 and 2) and a single default route



Used for stable-scalable internetworks



Minimizes the number of LSAs and the need for any external area SFP calculation

Not-so-stubby area (NSSA) 

Can import AS external routes and send them to other areas, but still cannot receive AS-external routes from other areas



Allows the injection of external routes in a limited fashion into the stub area

Cisco extension, although implemented by many vendors ! ! Totally Stubby area router ospf 100 area 20 stub no summary !

73

! ! Totally Stubby area router ospf 100 area 20 nssa !

Stub areas 

It is often used when an area has only one ABR 

In this case, we do not need to propagate external routes because there is only one path that connects to the rest of the network 



74

In fact, external routes in non-stub areas are useful only if more than one Egress Router exists

Stub areas are activated upon an explicit configuration coming from the network manager

Totally stubby areas: example 

OSPF does not prevent to have a totally stubby area with more than one Area Border Router 

Internal routers reach the external destinations based on the best ABR

Area 0 (backbone)

Area 0 (backbone)

R1

R2

R1

R3

Area 1 - TotStub

75

Area 0 (backbone)

R2

R1

R3

Area 1 - TotStub

R2

R3

Area 1 - TotStub

Stub/totally stubby areas: limitations 

A stub/totally stubby area can not be used to reach other domains 

It is thus not possible to place an AS Boundary Router into a stub area  

Let’s assume we have a Boundary Router into a Stub area If the packets that enter the area are directed to another area inside the AS, the Default Route is used 





The packets following the Default Route exit through the Boundary Router, that is to say through where they come from

It is forbidden to have Virtual Links through a stub/totally stubby area 

76

Only internal routes are known inside the stub area

More on Virtual Links later

Summary of LSA propagation

77

OSPF v3

78

Overview 

79

Defined originally for IPv6, in effect supports multiple protocols 

“Integrated routing” approach



Although many production networks rely on OSPFv2 for IPv4 and OSPFv3 for IPv6



Same mechanisms, but a major rewrite of the internals of the protocol

Main similarities and differences 

Main similarities 

General ideas are the same 



80

Neighbor discovery adjacencies

and

the

mechanism

for

managing



LSA flooding and related mechanisms



All kind of networks supported (point-to-point, NBMA, broadcast)

Main differences 

Clear distinction between “topology” and “network addresses”



All IPv4-specific semantics are removed



Carries IPv6 addresses



Link-local addresses used as source address in OSPFv3 packets



IPv6 transport: OSPF runs directly over the IPv6's network layer

OSPFv3 processing: per link, not per network 

This is due to the fact that OSPFv3 has a clear differentiation between topology (which is the same for all address families) and network addresses 

IPv6 uses the term link to indicate a communication facility or medium over which nodes can communicate at the link layer



OSPF interfaces connect to links instead of to IP subnets



OSPF for IPv6 therefore runs per-link instead of the IPv4 behaviour of per-IP-subnet, and the terms “network” and “subnet” are usually replaced by the term “link”



This change affects the receipt of OSPF protocol packets, and the contents of hello packets and network LSAs 



81

E.g., Hellov2 has a field named “network mask”, which is replaced by the “interfaceID” in Hellov3

Multiple instances per link supported

Use of link-local addresses 

82

Link-local addresses are used 

As source address in OSPF packets



As next hops in the routing table



OSPFv3 uses IPv6 link-local addresses to identify the OSPFv3 adjacency neighbours

Explicit support for multiple OSPFv3 instances 

83

Explicit support for multiple OSPFv3 instances per link 

Separate autonomous systems, each running OSPFv3, can use a common link



A single link could belong to multiple areas



OSPFv3 uses a new field, called the Instance ID, to allow multiple instances per link



To have two instances talking to each other, they must share the same instance ID



By default, the instance ID is set to 0

Use of IPv6 transport 

OSPFv3 uses IPv6 transport



Multicast addresses

84



FF02::5 —Represents all shortest path first (SPF) routers on the link-local scope, equivalent to 224.0.0.5 in OSPFv2



FF02::6 —Represents all designated routers (DRs) on the linklocal scope, equivalent to 224.0.0.6 in OSPFv2

OSPFv3 LSA: types 1-4 





Router LSA (type 1) 

As in OSPF v2, but no longer advertised adjacencies with IP stub networks



Used only to recreate the topology, no longer to describe the location of IP networks

Network LSA (type 2) 

As in OSPF v2, but no longer advertised adjacencies with IP network



Used only to recreate the topology, no longer to describe the location of IP networks

Summary-LSA (type 3) 



Summary LSA (Type-4) 

85

Renamed “Inter-Area-Prefix-LSAs”

Renamed “Inter-Area-Router-LSAs”

Inter-Area Prefix LSA (type 4): details 

Describes the destination outside the area but still in the AS



Summary is created for one area, which is flooded out in all other areas



Originated by an ABR



Only intra-area routes are advertised into the backbone



Link State ID simply serves to distinguish inter-area-prefixLSAs originated by the same router



Link-local addresses must never be advertised in inter-areaprefix-LSAs

86

Link LSA (type 8): new in OSPFv3 

Keeps the mapping between each link-local address and the IPv6 network it is attached to 

Surprising, it does not refer to an interface, but to an IPv6 network. In fact, each L-LSA is a tuple: Advertising router  Link-Local address  Network address 





87

E.g., 10.11.12.13 2001:1:1:3::/64



fe80::cc3b:edb2:29c6:9041



Has local-link flooding scope 

Never flooded beyond the link it is associated to, hence reaching only the neighbors of the router



Link local addresses are only used when calculating next hops during the OSPF routing calculation, hence they do not need to be flooded past the local link

Warning: having a link-local scope, the OSPF database is not exactly the same on all the routers of the area!

Inter-area prefix LSA (type 9): new in OSPFv3 

Inter-area prefix LSA (type 9): new in OSPFv3 

Contains the list of IPv6 networks each router R is attached to 





88

This link state includes a reference to the LSA we are referring to (i.e., Router LSA or Network LSA) Hence, each router can send multiple IA-LSA depending in its topology

Each DR will send an IA-LSA containing the IPv6 network it is serving

Link State in OSPFv3 (1)

NetA/PrefixA

NetA/PrefixA

Link local address mapping

IP6_global_1 IPv6_linklocal_1 R1

Link State Database: Link LSA (Adv Router: R1; #Prefixes: 1) Link local interface address: IPv6_linklocal_1; Address prefix: NetA

Inter-area prefix LSA (Adv Router: R1; #Prefixes: 1) Address prefix: NetA/PrefixA

89

Link State in OSPFv3 (2)

NetA/PrefixA

R1

R1

Link local address mapping

IP6_global_1 IP6_linklocal_1

Link local address mapping

IP6_global_2 IP6_linklocal_2 R2

NetA/PrefixA

R2

NetA/PrefixA

Link LSA (Adv Router: R1; #Prefixes:1)

Link LSA (Adv Router: R2; #Prefixes:1)

Link local address: IPv6_linklocal_1; Addr. prefix: NetA

Link local address: IPv6_linklocal_2; Addr. prefix: NetA

Router LSA (Adv Router: R1)

Router LSA (Adv Router: R2)

Neighbor RouterID: R2 - Point-to-point link to another router

Neighbor RouterID: R1 - Point-to-point link to another router

Inter-area prefix LSA (Adv Router: R1; #Prefixes: 1)

Inter-area prefix LSA (Adv Router: R2; #Prefixes: 1)

Address prefix: NetA/PrefixA

Address prefix: NetA/PrefixA

90

Link State in OSPF (3) R1 NetA/PrefixA

IP6_global_1 IP6_linklocal_1

Router ID R1

Router ID R2

Link local address mapping

Link local address mapping

NetA

(Transit network)

IP6_global_2 IP6_global_3 IP6_linklocal_2 IP6_linklocal_3

R2

R3

Link local address mapping

Router ID R3

Link LSA (Adv Router: R1; #Prefixes:1)

Router LSA (Adv Router: R1)

Link local address: IPv6_linklocal_1; Addr. prefix: NetA

Neighbor RouterID: R3 - Point-to-point link to transit network

Link LSA (Adv Router: R2; #Prefixes:1)

Router LSA (Adv Router: R2)

Link local address: IPv6_linklocal_2; Addr. prefix: NetA

Neighbor RouterID: R3 - Point-to-point link to transit router

Link LSA (Adv Router: R3; #Prefixes:1)

Router LSA (Adv Router: R3)

Link local address: IPv6_linklocal_3; Addr. prefix: NetA

Neighbor RouterID: R3 - Point-to-point link to transit router

Network LSA (Adv Router: R3; #Prefixes: 1)

Inter-area prefix LSA (Adv Router: R3; #Prefixes: 1)

Attached Routers: R1, R2, R3

Address prefix: NetA/PrefixA

91

Basic configuration differences (in Cisco) 

In OSPFv3, a routing process does not need to be explicitly created 



In OSPFv3, each interface must be enabled using commands in interface configuration mode 

92

Enabling OSPF for IPv6 on an interface will cause a routing process, and its associated configuration, to be created

This feature is different from OSPFv2, in which interfaces are indirectly enabled using the router configuration mode

Steps to Configure OSPF for IPv6 

Complete the OSPF network strategy and planning for your IPv6 network. For example, you must decide whether multiple areas are required



Enable IPv6 unicast routing 



ipv6 unicast-routing

Enable OSPFv3 on the interface 

ipv6 ospf area



(Optional) Configure OPSFv3 interface specific settings, including area, router priority, and OSPFv3 path cost



(Optional) Configure routing configuration mode, including summarization, and so on

93

specifics from router priority,

router route

Configuring OSPFv3 in Cisco IOS 

Similar to OSPFv2 



Interfaces configured directly 



Replaces network command

“Native” IPv6 router mode 

94

Prefixing existing Interface and Exec mode commands with “ipv6”

Not a sub-mode of router ospf

Configuration Modes in OSPFv3 

Entering router mode 



Entering interface mode 



95

[no] ipv6 router ospf

[no] ipv6 ospf area

Exec mode 

[no] show ipv6 ospf []



clear ipv6 ospf []

Cisco IOS OSPFv3 Specific Attributes 

Configuring area range 



96

[no] area range /

Showing new LSA 

show ipv6 ospf [] database link



show ipv6 ospf [] database prefix

OSPFv3 Debug Commands 





97

Adjacency is not appearing 

[no] debug ipv6 ospf adj



[no] debug ipv6 ospf hello

SPF is running constantly 

[no] debug ipv6 ospf spf



[no] debug ipv6 ospf flooding



[no] debug ipv6 ospf events



[no] debug ipv6 ospf lsa-generation



[no] debug ipv6 ospf database-timer

General purpose 

[no] debug ipv6 ospf packets



[no] debug ipv6 ospf retransmission



[no] debug ipv6 ospf tree

Enabling OSPFv3 on an interface 

Most of the OSPFv3 configuration is done on the interface

Router# configure terminal Router(config)# interface ethernet 0/0 Router(config-if)# ipv6 address 2001:400:1::1/64 Router(config-if)# ipv6 ospf 1 area 0 Router(config-if)# ipv6 ospf 1 cost 20

98

Configuring OSPFv3 routing specifics (1) 

OSPFv3 requires general routing specifics commands to be configured from router configuration mode 

Not in interface mode



Entering in OSPFv3 global configuration mode with 

99

ipv6 router ospf PID

Configuring OSPFv3 routing specifics (2) 

Basically, only one information required here: the RouterID 

Unfortunately, OSPFv6 uses still a 32bit RouterID 







The “ipv6 router-id ” command compulsory when an IPv6-only router is deployed

becomes

The OSPFv3 router ID can be expressed in dotted decimal, allowing easy overlay of an OSPFv3 network on an existing OSPFv2 network

If IPv4 is configured on the router, by default, the router ID is chosen in the same way as it is with OSPFv2 



100

And, even more unfortunately, is not able to set that number automatically, when no IPv4 addresses are available

The highest IPv4 address configured on a loopback interface becomes the router ID If no loopback interfaces are configured, the highest address on any other interface becomes the router ID

Configuring OSPFv3 routing specifics (3) 

Example

Router# configure terminal Router(config)# ipv6 unicast-routing

Router(config)# ipv6 router ospf 1 [*] Router(config-rtr)# router-id 2.2.2.2 [*] Upon issuing that command, the router prints the following text on the console: Router(config-rtr)# *Mar

1 00:00:52.575: %OSPFv3-4-NORTRID: OSPFv3 process 1 could

not pick a router-id, please configure manually Router(config-rtr)#

101

OSPFv3 Route Summarization Before Summarization:

After Summarization:

102

Conclusions 

Very powerful (and very complex) protocol



Widely used in modern networks

103