Link State & OSPF Spring 2013CE 151 - Advanced Networks1

CE 151 - Advanced Networks 1

Link State & OSPF

Spring 2013


Administrativia• Static Routing lab

– Wording problems– VM stability issues

• How are the labs going?– I will start upgrades of VMs today… will send e-mail w/ new password when done

• Opportunities– Cruzio… I’m waiting to hear back– NMO Software Development for Cisco Advanced Services… waiting for applications– Expect more from campus network operations group…

• Next week– Link Layer lab due Wednesday, 4/24– Link-State Routing quiz Thursday, 4/25

• Project proposal due Tuesday 4/30– Topic– Draft outline– What you need to investigate

Spring 2013


Routing Introduction

• Remember… delivery of IP packets implemented by two processes– Forwarding– Routing

• Dynamic routing process is a distributed computation– Triggered by topology changes– Processing defined by a routing protocol– Output of the computation is forwarding state

• Goal of computation is convergence– After finite sequence of topology changes– Process should terminate updates to forwarding state– Forwarding state should be correct

• Loop-free (after convergence vs. instantaneous)• With desired characteristics: performance, possibly conform to policy

Spring 2013


“Fate-sharing” Principle

• “The fate-sharing model suggests that it is acceptable to lose the state information associated with an entity if, at the same time, the entity itself is lost.” Dave Clark, “Design Philosophy of the DARPA Internet Protocols”, SIGCOMM ‘88.

• Benefits– Ensures the failure of any single component of an internet does not invalidate

state located elsewhere in the internet– Localizing the effects of any failures– More robust system

• Internet routing architecture co-locates– Forwarding state– Routing process that computes the state

• Compared with virtual-circuit routing?

Spring 2013


Routing Protocol• A set of algorithms and messages that are used to exchange topology information

and populate the forwarding table with the routing protocol’s choice of best paths.

• Purpose:– Discover remote subnets– Maintain up-to-date forwarding tables– Choose the best path to destination subnets

• Components of a routing protocol:– Algorithm:

• Procedures for– Processing routing information – Selecting best-paths

• The data structures needed for these steps.

– Routing protocol messages: • Discover neighboring routers• Exchange topology information

Spring 2013


Classifying Routing Protocols

• Function:– Intra-domain/Interior Gateway Protocol (IGP)– Inter-domain/Exterior Gateway Protocol (EGP)

• Algorithm… distinguished by information exchanged:– Distance-Vector– Link-State– Path-Vector

Spring 2013


Functional Classification

• An autonomous system (AS) or routing domain is a region of the Internet that is administered by a single entity

– UCSC’s network– IBM’s corporate network– AT&T’s ISP network

• Routing inside an AS– Focus is on performance– Popular protocols: RIP, OSPF, IS-IS– Called intra-domain or internal gateway (IGP) routing

• Routing between ASs– Focus is on policy– Popular protocol: BGP– Called inter-domain or external gateway (EGP) routing

Spring 2013

CE 151 - Advanced NetworksSpring 2013 8

How ensure correct routes?

• Recall requirement for correctness of routing protocol– Loop-free

– Desired path characteristics

• Two strategies for ensuring correctness– Use identical algorithm for selecting paths

• Share minimal topology information

• Use identical path selection algorithm at all nodes

• Used for IGP/Intra-domain routing

• Use link-state or distance vector protocol

– Use custom (private) algorithm for selecting paths• Share full path information

• Use policy-specific path selection algorithm at each node

• Used for EGP/Inter-domain routing

• Use path-vector protocol


Algorithm Classification• Distance-Vector

– Vectors of destination and distance sent to neighbors• “Tell your neighbors about the rest of the network”

– Destination in terms of a network prefix– Distance in terms of a metric: hop count, delay, bandwidth– Use Distributed Bellman-Ford path selection algorithm– Popular protocol: Routing Information Protocol (RIP)

• Link-State– Flood description of your links (link state)

• “Tell the rest of the network about your neighbors”

– Links described by• End-point routers of subnet in internet• Cost of subnet: delay, bandwidth

– Use Dijkstra path selection algorithm– Popular protocol: Open Shortest Path First (OSPF)

• Path-Vector– Routes advertised as full-paths– Paths described by sequence of ASs– Popular protocol is Border Gateway Routing Protocol (BGP)

Spring 2013


Destination-Based Forwarding• Internet routing uses a single path per destination

• Destination-based forwarding is a restricted version of single-path

• A path through a node to a destination…

• …must be an extension of the path from the node to the destination.

• This is coming back to haunt us…

Spring 2013

CE 151 - Advanced Networks

Review

• The Internet implements a distributed routing architecture that is triggered by

topology change events.

• Convergence of routing in the Internet depends on the stability of the network

topology for a sufficient period of time.

• The Fate-Sharing Principle – “It is acceptable to lose the state information

associated with an entity if, at the same time, the entity itself is lost.”

– Ensures the failure of any single component of an internet does not invalidate state

located elsewhere in the internet

– Localizing the effects of any failures

– Results in a more robust system

– Achieved in Internet by co-locating

• Forwarding state

• Routing process that computes the state

11Spring 2013

CE 151 - Advanced Networks

Review

• An Autonomous System (AS) is a region of the Internet that is administered

by a single entity and follows a single routing policy.

• Functional classification of routing protocols

– IGP - routing inside an AS

• Share minimal topology information

• Use identical path selection algorithm at all nodes

• Optimize performance

– EGP - routing between ASs

• Share full path information

• Use custom path selection algorithm at each node to implement desired policies

• Enforce policies

12Spring 2013


Review

• Algorithmic classification of routing protocols

– Distance-vector

• Send vectors of distances to destinations to neighbors

• “Tell your neighbors about the rest of the network”

• “Distributed computation”

– Link-state

• Flood description of your links to all routers


• “Distributed database”

– Path-vector

• Distance-vector with full paths

Spring 2013


Review

• Destination-based forwarding– A path through a node to a destination… must be an extension of the path from

the node to the destination– Tends concentrates traffic on a subset of the network topology.

Spring 2013


Link-State

Spring 2013


Routing Algorithms• Distance-Vector








Spring 2013


Routing Algorithms• Distance-Vector








Spring 2013


How ensure correct routes?• Recall requirement for correctness of routing protocol

– Loop-free– Desired path characteristics


• Share minimal topology information• Use identical path selection algorithm at all nodes• Used for IGP/Intra-domain routing• Use link-state or distance vector protocol

– Use custom (private) algorithm for selecting paths• Share full path information• Use policy-specific path selection algorithm at each node• Used for EGP/Inter-domain routing• Use path-vector protocol

Spring 2013


How ensure correct routes?• Recall requirement for correctness of routing protocol

– Loop-free– Desired path characteristics


• Share minimal topology information• Use identical path selection algorithm at all nodes• Used for IGP/Intra-domain routing• Use link-state or distance vector protocol

– Use custom (private) algorithm for selecting paths• Share full path information• Use policy-specific path selection algorithm at each node• Used for EGP/Inter-domain routing• Use path-vector protocol

Spring 2013


Link-State Protocols• Are Interior-Gateway Protocols (IGPs)

• Exchange link-state information– Pair of routers connected by a subnet– Cost of subnet (hop count, delay, etc.)

• Conceptually, very simple…

Spring 2013


Link-State Protocols• Maintains a topology database of all the links it has heard of

– Initialize with the subnets it is connected to.

• Floods link-state updates describing its directly connected subnets, including any changes to these links.– “Tell the rest of the network about your neighbors”

• Participates in the flooding of link-state updates from other routers.

• On update of its topology database– Runs a shortest-path algorithm on the database to compute routes

• Dijkstra is most efficient– Updates its forwarding table with any changes.

Spring 2013


Characterizing Link StateLink-State• # updates per link change?

– One.

• How far propagate updates?– Flooded to all nodes.

• One update, global distribution.

• Scaling problems due to flooding

• As we’ll see next lecture, the characteristics of distance vector…– …are very different– …hint at a much better solution

Spring 2013


Dijstra Shortest-Path Algorithm• Breadth-first search of paths, by increasing path cost, for best paths to all

destinations. Terminate when path has been found for all destinations.

• Maintain two sets– Destinations for which shortest paths have been found.

• Permanently labeled destinations P• Initialize with self

– Destinations for which candidate shortest paths have been found.• Temporarily labeled destinations T• Initialize with my neighbors.

• Iterate– Move shortest path in T, say for destination D, to P– Add routes for D’s neighbors, that are extensions of the path to T, to T if they are shorter

than the current path in T for each neighbor. The “relaxation” step.– Repeat until a route has been added to P for all destinations

Spring 2013


More formally…• E is the set of edges.

• wij is the weight of the link between nodes i and j.

• P and T… see previous slide.

• P and T entries are triples, <d, p, w>:– d is the destination– p is the predecessor– w is the link weight

Spring 2013


Dijkstra… path costC,B,2

G, H,6

2 2

2F,E,1A,0 D,

73

2

32

4

C,

F,E,4A,0

B,2

D,

G, H,C,

F,E,4A,0

B,2

D,

G,5 H,

C,

F,6E,4A,0

B,2

D,

G,5 H,C,

F,6E,4A,0

B,2

D,

G,5 H,8

C,9

F,6E,4A,0

B,2

D,

G,5 H,8C,9

F,6E,4A,0

B,2

D,10

G,5 H,8Spring 2013


Translating to a Protocol• Dijkstra requires a “centralized” implementation

– Maintain a full graph of the network, on an event-driven basis– Re-compute routes as the graph changes– Flood changes to your links

• “Brute-force” protocol.– Straight-forward, easy to understand– Inefficient… lots of overhead

Spring 2013


Review• Dijkstra

– Iterates on “next shortest path”– Requires centralized computation

• LS protocols– Is an IGP– Implements a centralized routing model– Floods link-state updates describing current state of its links

• “Tell the rest of the network about your neighbors”– Use Dijkstra algorithm because it is most efficient shortest-path algorithm

Spring 2013


The Challenge of Internet Routing• Independent routing computations at each router…

• Need to compute paths that…– Support destination-based forwarding– Are shortest– Are loop-free

• This is trickier than it looks…

Spring 2013


Examples illustrating challenges of distributed routing…

Spring 2013


Correct solution depends on combination of routing algorithm and algebra used for metrics.

Spring 2013


OSPF

Spring 2013


OSPF• OSPF = Open Shortest Path First• The most widely used routing protocol• The complexity of OSPF is significant

• History:– 1989: RFC 1131 OSPF Version 1 – 1991: RFC1247 OSPF Version 2– 1994: RFC 1583 OSPF Version 2 (revised)– 1997: RFC 2178 OSPF Version 2 (revised)– 1998: RFC 2328 OSPF Version 2 (current version)

Spring 2013


What We Cover…

• Messages

• Router IDs

• Flooding process

• Metrics

• Designated routers

• Areas

Spring 2013


OSPF Messages• An OSPF message can contain one of five packet types.

• OSPF is embedded directly in an IP frame (doesn’t use UDP)– Protocol field is set to 89 (OSPF)– Destination address is typically set to one of two multicast addresses:

• 224.0.0.5 (“AllSPFRouters”… Hello messages) or • 224.0.0.6 (“AllDRouters”… routing info to “Designated Routers”).

• If the OSPF packet is encapsulated in an Ethernet frame, the destination MAC address is also a multicast address: – 01-00-5E-00-00-05 or 01-00-5E-00-00-06

Spring 2013


OSPF Packet Types• Hello: Used to establish and maintain adjacency with other OSPF routers.

• DBD: The database description (DBD) packet contains an abbreviated list of the sending router’s link-state database and is used by receiving routers to check against the local link-state database.

• LSR: Receiving routers can then request more information about any entry in the DBD by sending a link-state request (LSR).

• LSU: Link-state update (LSU) packets are used to reply to LSRs and to announce new information. LSUs contain seven different types of link-state advertisements (LSA).

• LSAck: When an LSU is received, the router sends a link-state acknowledgment (LSAck) to confirm receipt of the LSU.

Spring 2013


OSPF Message Header

2: current version is OSPF V2

Message types:1: Hello (tests reachability)2: Database description3: Link Status request4: Link state update5: Link state acknowledgement

ID of the Area from which the packet originated

Standard IP checksum taken over entire packet

0: no authentication1: Cleartext password2: MD5 checksum(added to end packet)

Authentication passwd = 1: 64 cleartext password Authentication passwd = 2: 0x0000 (16 bits)

KeyID (8 bits) Length of MD5 checksum (8 bits) Nondecreasing sequence number (32 bits)

Prevents replay attacks

Spring 2013


RouterIDs

Router ID

Router ID

Router ID

Router ID

Router ID

Router ID

• Router ID plays an important role in OSPF– Uniquely identifies each router in a routing domain– Used in Designated Router election process

(explained later)

• Router ID is an IP address of a router

• Cisco routers use following algorithm to determine Router ID

– IP address configured with OSPF router-id command

– If not configured, use highest IP address of a loopback interface

– If no loopback interfaces, use highest active IP address of physical interface

• Advantage of loopback interface is it cannot fail

Spring 2013


OSPF Hello Packets• Used to

– Discover OSPF neighbors and establish neighbor adjacencies– Elect the Designated Router and Backup Designated Router on

multiaccess networks such as Ethernet and Frame Relay– Negotiate

• Hello interval (e.g. 10 sec on Ethernet segments)• Dead interval: time to declare neighbor down (4x Hello interval)• Network type

• Five network types– Point-to-point– Point-to-multipoint– Broadcast multiaccess (Ethernet)– Nonbroadcast multiaccess (Frame Relay)– Virtual links

Spring 2013


OSPF LSU Packets• Link State Update (LSU) packets

– Used for OSPF routing updates– Contain one or more LSAs

• Link State Advertisements (LSAs)– Contain route information for destination networks– There are 11 types of LSAs

Spring 2013


Link State Advertisement (LSA)• The LSA of router 10.10.10.1 is as follows:

• Link State ID: 10.10.10.1 = can be Router ID• Advertising Router: 10.10.10.1 = Router ID• Number of links: 3 = 2 links plus router itself

• Description of Link 1: Link ID = 10.1.1.1, Metric = 4• Description of Link 2: Link ID = 10.1.2.1, Metric = 3• Description of Link 3: Link ID = 10.10.10.1, Metric = 0

Spring 2013

Each router sends its LSA to all routers in the network(using a method called reliable flooding)


LSA Format

Spring 2013

LSA Header

Link 1

Link 2


OSPF Metrics• The OSPF metric is called cost. The following passage is from RFC 2328:

– A cost is associated with the output side of each router interface. This cost is configurable

• RFC 2328 does not specify which values should be used to determine the cost.

• In Cisco IOS – Link cost is 108 ÷ link bandwidth.– Reference bandwidth (108) can be changed with auto-cost referencebandwidth– Modify link bandwidth value with bandwidth

Spring 2013


Link State Database• The collection of all LSAs is called the link-state database

• Each router has and identical link-state database– Useful for debugging: Each router has a complete description of the network

• If neighboring routers discover each other for the first time, they will exchange their link-state databases

• The link-state databases are synchronized using reliable flooding

Spring 2013


Link State Database

Spring 2013

Each router has a database which contains the LSAs from all other routers


Neighbor Discovery• Router multicasts OSPF Hello packets on all OSPF-enabled interfaces.• If two routers share a link, they can become neighbors, and establish an

adjacency

• After becoming a neighbor, routers exchange their link state databases

Spring 2013

Scenario:Router 10.1.10.2 restarts


Neighbor discovery and database synchronization

Spring 2013

Scenario:Router 10.1.10.2 restarts

Sends empty database description

Discovery of adjacency

Sends database description. (description only contains LSA headers)

Database description of 10.1.10.2Acknowledges

receipt of description

After neighbors are discovered the nodes exchange their databases


Regular LSA exchanges

Spring 2013

10.1.10.2 explicitly requests each LSA from 10.1.10.1

10.1.10.1 sends requested LSAs 10.1.10.2 has more

recent value for 10.0.1.6 and sends it to 10.1.10.1(with higher sequence number)


Dissemination of LSA-Update• A router sends and refloods LSA-Updates, whenever the topology or link

cost changes. (If a received LSA does not contain new information, the router will not flood the packet)

• Exception: Infrequently (every 30 minutes), a router will flood LSAs even if there are no new changes.

• Acknowledgements of LSA-updates:– explicit ACK, or– implicit via reception of an LSA-Update

Spring 2013


Why Designated Routers?• Large number of adjacencies

– Full mesh of adjacencies– n(n - 1)/2 adjacencies

• Excessive load from flooding LSAs

Spring 2013


Electing Designated Router• OSPF elects a Designated Router (DR) on multiaccess networks• DR is collection and distribution point for LSAs on network• Backup Designated Router (BDR) also elected for case where DR fails

Spring 2013

224.0.0.6

224.0.0.5


Electing Designated Router• Election held through exchange of Hello messages

• DR/BDR election criteria– DR is router with highest interface priority– BDR is router with second-highest interface priority– If interface priorities are equal, use Router ID

Spring 2013


Electing Designated Router• Further elections only occur on failure of DR/BDR

• On DR failure– BDR promoted to DR– New BDR elected

• On BDR failure– New BDR elected

• Live (B)DR not replaced by election.– If DR dies

• BDR replaces DR• Election held to replace BDR

– If BDR dies• Election held to replace BDR

Spring 2013


OSPF Areas• An OSPF Area is a set of routers that share link state information

• Goal is to address scalability problem with OSPF flooding

• Area 0 is the backbone area– All areas must be connected to he backbone area– Best practice to use area 0 in single-area OSPF– Eases conversion to multi-area

Spring 2013


Review• OSPF message types

– Hello - establish and maintain adjacency with OSPF routers– DataBase Description (DBD) - summary of my database– Link-State Request (LSR) - request for detailed entry– Link-State Update (LSU) - response to LSR– Link-State Ack (LSAck) - acknowledge receipt of LSU

• Router IDs…– Uniquely identifies each router in a routing domain– Used in Designated Router election process – Best practice is to use loopback interface... it can’t fail

Spring 2013


Review• OSPF metrics…

– “Cost is associated with the output side of each router interface.”– In IOS is 108 / link bandwidth (inversely proportional to bandwidth)

• Designated routers…– Mitigate impact of “full-mesh” on # of adjacencies in topology (n(n-1)/2)– DR is collection and distribution point for LSAs on network

• OSPF deals with scaling problem by using “areas,” area 0 is the backbone.

Spring 2013

Documents

Link State & OSPF Spring 2013CE 151 - Advanced Networks1