Routing Notes by Keith 

Topic Links 

Administrative Distances

Metrics

Classful and Classless Protocols

Common Ports

Routing Algorithms and Features

OSPF

Queuing Strategies

EIGRP  

Administrative Distances

An administrative distance is a measure of the trustworthiness of a route.The router uses this to determine the best path when there is information from more than one routing protocol.The lower the administrative distance the greater the reliability of the information. The EIGRP summary route can only be seen on the table where the summary was configured. 

  Routing Metrics

Each routing protocol uses different methods to determine the best path.In order to determine the best path a routing metric is created.Depending on the protocol it may be as simple as hop count (RIP v1) or much more complex as is the case with IGRP which uses up to 5 factors in determining the final metric.The standard metric formula for IGRP is:

 

Metric = (K1*Bandwidth) + [(K2 * bandwidth) / (256 – Load)] + (K3 * Delay) Then

IF K5 {MTU} is not zero, Metric = Metric * (K5/ (reliability + K4))

 

Given default constant values: Metric = bandwidth + delay

 

Bandwidth is in Kbps and delay is in microseconds and can be determined with the show interfaces command.

 

 

Classfull and Classless Protocols

 

IGRP and RIP v1 are Classfull protocols and will only do route summarization at the class boundaries.

OSPF, EIGRP, RIP v2, BGP-4, and IS-IS are classless protocols.Route summarization is manual and selected boundaries.

 

Classless protocols use different subnet masks within a network, this is called variable-length subnet masking or VLSM.This permits more selective summarization that would be possible with Classfull routing.

 

 

Common Ports

 

TCP – Port- 6UDP– Port-17TFTP - 69

RIP – Port-520IGRP –Protocol 9

DNS – Port-53SNMP – Port- 161

 

Routing Algorithms and Features

 

RIP and IGRP use Bellman-Ford for route calculation

EIGRP uses Diffusing Update Algorithm (DUAL)

 

Distance Vector Protocols

 

Feature	RIPv1	RIPv2	IGRP	EIGRP
Count to Infinity	YES	YES	YES	NO
Split Horizon	YES	YES	YES	YES
Hold Down Timer	YES	YES	YES	NO
Triggered Updates w/ Rt Poisining	YES	YES	YES	YES
Load Bal Equal Paths	YES	YES	YES	YES
Load Bal UnEqual Paths	NO	NO	YES	YES
VSLM Support	NO	YES	NO	YES
Metric	HOPS	HOPS	Composite	Composite
Hop CountLimit	15	15	100	100

 

 

The default for IGRP and EIGRP on hop count limit is 100 but it can be configured to be 255.For IGRP that makes it

scalable for medium size enterprises and for EIGRP for large enterprises.

 

Link State Protocols

 

Feature	OSPF	IS-IS	EIGRP
Hierarchical Topology Required	YES	YES	NO
Retains Information on all possible routes	YES	YES	YES
Route Summarization (Manual)	YES	YES	YES
Route Summarization (Automatic)	NO	NO	YES
Event-Triggered Announcements	YES	YES	YES
Load Balancing (Equal-Paths)	YES	YES	YES
Load Balancing (Unequal Paths)	NO	NO	YES
VSLM Support	YES	YES	YES
Routing Algorithm	Dijkstra	IS-IS	DUAL
Hop-Count limit	Unlimited	1024	100

 

 

 

 

 

OSPF

 

 

1)      General Basics of OSPF

2)      Database Types

3 Multicast Addresses

4)      Hello Packet Structure

5 Election and Exchange Process

6)      Modes and Topologies

7)      Basic OSPF Commands

8)      Description of Update Process

9)      Keys to Understanding OSPF in Larger Network

 

 

 

Open Shortest Path First version 2 (OSPF) is:

an Interior Gateway Protocol (IGP)

is described inRFC 2328

has faster convergence than RIP

floods routing changes throughout network

supports Variable length subnet masks (VSLM)

 

OSPF was written for larger networks, specifically 50 + routers.

 

It uses protocol number 89 with:

6- TCP

17- UDP

 

OSPF uses cost metrics assigned to the interface output side, called:

interface output cost and based on

the speed of the media (bandwidth)

 

OSPF Routers form adjacencies with neighbors.To have an adjacency means

that the databases for the two routers have been synchronized.

 

Link States are advertised to other routers with LSAs

Link State Advertisements

 

Neighbors are defined as two routers that have interfaces on a

common network.These are discovered and maintained with

Hello packets (TOP)

 

OSPF has three databases:

Neighbor database – bi-directional communication

Link-State database – (topology database)

Routing table - (forwarding database) which is created using the

SPF algorithm (Dijkstra algorithm)

 

OSPF Topologies

Broadcast Multiaccess

Point-to-Point

Nonbroadcast multiaccess (NBMA) [Frame or X.25] (TOP)

 

Hello packets are sent out to an address called the AllSPFRouter address as

multicast on 224.0.0.5 on MAC address: 010005E 0000005 it also uses

multicast on 224.0.0.6 on MAC address: 010005E 0000006 (TOP)

 

Hello Packets include:

 

Router ID:This is the highest IP address on an active interface

Used to break ties in DR and BDR elections if priority is tied

Hello and

dead intervals:Time between sends (10 sec default) and the time before a link is considered down is the dead interval, generally 4 time the hello interval. (NBMA is 30 seconds and 120 seconds respectively.)

 

Neighbors:Has bi-directional communication meaning that the router sees itself in the other routers hello packet.

 

Area ID:Routers that share a common segment have the same subnet and mask and have the same link-state information.

 

Router Priority:8-bit number that indicates the priority of the router during DR and BDR elections.The higher priority the more likely the selection. Each link has its own election process set at the interface configuration mode.

BDR and DR

IP Addresses:The addresses of the BDR and DR

 

Authentication

PasswordSet if authentication is used

 

Stub area flag:Two routers must agree on this in hello packets

 

 

OSPF Packet Header Includes:

 

Version Number

Type

Hello

Database Description (adjacency)

Link-state request

Link- state update

Link-state Ack

Packet Length

Router ID (source)

Checksum

Authentication Type:

0 – no authentication

1 – clear text

2- MD5

Authentication Information

Data – routing information

 

On multiaccess networks the DR and BDR get routing updates but only the DR sends out

the updates, unless the DR goes down.Each router must set up an adjacentcy with the DR and the BDR.After the initial election new routers form an adjacency with the DR and the BDR. (TOP)

 

 

Election Process:

Router with highest priority is the DR and the next highest is the BDR

The default interface priority is 1.

The highestrouter ID breaks ties.

A router with the priority of 0 cannot be a BDR or DR.

A router that is not the DR or BDR is called a Drother.

If a DR goes down, the BDR becomes the DR and a new BDR is elected

Higher priority routers added to the network do not cause a new election.

BDR uses a reliability timer – if no LSA is heard then it assumes the DR is down.

 

OSPF Exchange/Startup Process Upon All Routers Coming Up at the Same Time

Step 1:Router is in down state and sends a hello packet our all participating interfaces on 224.0.0.5.

Step 2: Router that get packet from (1) add the sending router to their list of

neighbors. This is the init state.

Step 3: All routers that got the first packet send a reply hello packet that includes

all other neighboring routers.

Step 4: The first router sees its own ID in the packet and adds the other routers to its neighbor database. This creates the two-way state.

Step 5: Election for DR and BDR is held. Once this is held the routers are in an

exstart state where the exchange protocol movesthe routers toward a full state. With the exchange protocol the DR and the BDR establish adjacencies witheach router on the network, then based on the higher router ID they decide on a master/slave relationship with the master as the higher router ID.Link state info is exchanged between the DR and BDR and the router with which they have adjacencies.Database Description Packets (DBDs or DDPs (Database Descriptor Packets)) are exchanged with sequence numbers defined by the master.The slave gets the DBD, sends a LSAck, compares the DBD to see if it is more up to date. If it is more up to date, it asks for a Link-State Update (LSU) by sending a Link-State Request(LSR). This is called the Loading state. When all the LSRs are satisfied, the routers are in the full state.

Step 6: Hello packets are sent each 10 sec. including the names of the DR and BDR.

 

OSPF works on cost metrics, the higher the bandwidth, the lower the cost.

 

OSPF keeps up to 6 equal cost route entries in the table for load balancing.

The default is 4 but the maximum-paths router configuration command.

will allow up to 6. paths to same destination.

 

Flapping causes new LSU’s, so each time a LSU is received the router waits for a

period of time before recalculating the routing table (5 seconds is the default.) the

times spf spf-delay spf-holdtime command allows this to be configured.

 

The SPF algorithm forms an SPF Tree which is a loop free shortest path to all

networks with the router as the root.

 

MultiaccessLink state process

 

Step 1:Router notices a change floods it to 224.0.0.6 the all DR and BDR address with an LSU packet that includes one or many LSAs.

Step 2:The DR acknowledges this and floods the LSU to the other routers on the network on 224.0.0.5.Each router responds with an LSAck.

Step 3 : If the router is connected to other networks it forwards the LSU to the DR of those networks or to the adjacent routers if point-to-point.Those DRs then multicast the LSUs

Step 4: When a router receives the LSU is takes the changed LSA’s and updates its link-state database. This is then used to run the SPF algoithm and update the routing database.

 

LSA age field – aging timer = default of 30 minutes.When this expires the router that originated the LSU sends an update to say the link is still valid.

 

Entries that already exist are discarded.

 

Point-to-Point links have no election. Both routers automatically become adjacent

routers.

 

debug ip ospf adj -Lookat the point-to-point adjacency election

 

For BRI/PRI and Asynchronous

BRI/PRI – use dialer-map along with OSPF configuration + Broadcast to indicate that broadcasts should be forwarded to the protocol address.

For Asynchronous use:

async default routing (TOP)

 

 

Non-Broadcast MultiAccess Topologies (NBMA)

1. Broadcast option must be enable for all Virtual Circuits

2. Full mesh – (n(n-1))/2 where n is the number of sites

 

Two modes of OSPF in NBMA Networks

Non-Broadcast MultiAccess – All broadcast packets are replicated and sent to all routers, usually in a fully meshed topology.Possible configuration to be sure all adjacencies are set. RFC-defined

 

Point-to-Multipoint – treats the NBMA network as a set of point-to-point links, no election, usually used in partially meshed networks. RFC –defined.

 

NBMA topology often uses subinterfaces.

Router(config)# interface serial number.subinterface-number {multipoint | point-to-point}

 

Default OSPF mode for point-to-point subinterface is point-to-point mode

Default OSPF mode for point-to multipoint subinterface mode is non-broadcast MultiAccess mode.

 

In NBMA mode sometimes a subinterface can fail while the main link or keep alives from another subinterface continue meaning that OSPF fails to notice the down link.

 

Point-to-multipoint does not require full mesh, but all routers are on one IP subnet. It does not require the static neighbor configuration.

 

Other Modes:

Point-to-multipoint non-broadcast requires static definition of neighbors

Broadcast mode – adjacency is automatic – Cisco standard

Point-to-point NBMA mode –(TOP)

 

 

OSPF Commands

router ospf process-id

router(config-router)# network address wildcard-mask area area-id

 

{note on wildcard masks – 1 means don’t care and 0 means match}

 

To set a higher router ID use loopback command:

 

router(config)# interface loopback number

 

{Since loopback addresses are always active a loopback address can be more reliable for key routers}

 

To determine the router ID of a router:

 

show ip ospf interface

 

To change the ospf priority on a router:

 

Router(config-if) ip ospf priority number (Can be 1-255)

 

The highest priority is the DR, 1 is the default and 0 is a drother.

 

To change the link cost:

 

ip ospf cost cost

 

Some default costs are:

56 kbps serial link – 1785

T1 - 64

Ethernet10

16 Mbps Token Ring6

 

To change the cost use the:

 

router(config-router) #auto-cost reference-bandwidth reference-bandwidth

 

The reference bandwidth is 100Mbps or 10^8 bps.Also ip ospf cost overrides the

calculation process.

 

Setting mode type:

 

router(config-if)# ip ospf network command-mode

nonbroadcast

point-to-multipoint

point-to-multipoint non-broadcast

broadcast

point-to-point

 

The neighbor command is used to configure ospf neighbors:

 

router(config-router)#neighbor ip-address [priority number] [poll-interval sec] [cost number]

 

Verification commands

show ip route

show ip protocols

show ip route ospf

show ip ospf interface e0

show ip ospf

show ip ospf neighbor

show ip ospf neighbor detail

show ip ospf database

clear ip route * (resets routing table)

clear ip route destination network

 

debug ip ospf events – displays information about protocol related events such as flooding, DR/BDR election and spf calculations

 

debug ip ospf packet – displays information about each OSPF packet received

 

Other parameters for debug ip ospf include:

Adjacent

flood

lsa-generation

retransmission

spf

tree

(TOP)

 

 

Description of update process:

 

A router receives an LSA and first checks if the LSA is from an external network or if

the router itself is a stub router.If either of these items is true then the router acknowledges

the LSA then discards the LSA.Next the router checks if the LSA timer (MAXAGE) has

expired or if the neighbor is in a loading or exchange state.If either of these items is true

then the LSA is acknowledged and discarded.If this false, then the router check to see if

the LSA is in its topological database.If it is in the topological database and the LSA received

is more recent than the one in the database then a new LSA is sent to the sender of the old LSA,

and then the arrival of the LSA is checked against the last run of the SPF algorithm and if the

minimum timer after the last run of the algorithm has expired, then the LSA is flooded out all

interfaces except for the arriving interface.If the LSA is less recent than LSA already in the

database, then the packet is discarded and acknowledged.The more recent LSA is installed

in the topological database, time stamped with the arrival time and acknowledged.

 

(TOP)

 

OSPF Larger Networks

 

For an OSPF network with Multiple Areas there are a set ofkey concepts:

 

AREA 0 – This is the backbone area.

ABRs - Area Border Routers

ASBRs - Autonomous System Border Routers

Stub Areas -Only accept internal route updates

Totally Stubby Areas -Only accepts default route to other networks

Not So Stubby Areas - Accepts limited summary routes to other networks

LSA Types - Internal and External LSAs are key to working with area types.

 

 

Summarization at area boundaries and the definition of area types control the size of routing tables.

 

Queuing Strategies

 

1.)First In First Out – (FIFO) – This is the default queuing method for all interfaces over E1(2.048 Mbps).Packets come in and they are buffered and go out in the order in which they arrived. This is the fastest method but makes no distinction between types of packets.

 

2.)Priority Queuing – Each packet type is assigned a priority of high, medium, normal or low.The high queue is emptied first then medium, normal then low. It is possible that lower priority traffic can never get sent.

 

3.)Custom Queuing – 16 queues are allowed and bandwidth assigned to each.More sensitive traffic an be assigned larger bandwidth.

 

4.)Weighted-Fair Queuing – A complex algorithm determines the distribution of packets assigning them precedence and breaking up large packet streams so all traffic can get through. This is the only dynamic queuing method.

 

 

 EIGRP - is a mixed distance vector/link state routing protocol.  It is proprietary for Cisco. There are two versions of EIGRP, v1 and v2  which has been available since 11.1(3).  v2 includes many enhancements that aid in its stability.

 

        Key advantages are:

            Rapid convergence

            Only network change information is propagated

            Normal operation in a stable network yields only hello traffic

 

        EIGRP uses DUAL (Diffused Update Algorithm) to determine the best path.

 

        EIGRP chooses the best path as the successor path and next best path as the feasible successor. Knowledge about the network is derived from hello packets sent on a 5 second basis for high bandwidth links and every 60 seconds for lower bandwidth links.  These hellos generate neighbor information.    When a router sees another router's hello packets it becomes a neighbor.

   

        Examples of low bandwidth circuits are multipoint frame T1 or less circuits or ISDN BRI circuits.

 

        Hello Interval - the time between hello packets

        Hold Time - the amount of time where a router does not receive a hello packet.  This is usually 3 times

                            the Hello interval.  By default this would be 15 and 180 seconds.

 

        The hello interval can be adjusted with the ip hello-interval eigrp

        The hold time can be adjusted with the ip hold-time eigrp

        To see eigrp neighbors type - show ip eigrp neighbor

 

        Eigrp does not build peers over secondary addresses.

        In a point to multi-point topologies the broadcast key work must be used in the frame-relay map command.

 

        Eigrp installs routes in a topology table which can be seen with the show ip eigrp topology

        statement.  This table has the information needed to build a set of vectors and distances needed to reach each

        network.

 

        EIGRP Metrics:

 

        The base formula for EIGRP metrics is:

   

        metric = [K1 * bandwidth + (K2 * bandwidth) / (256 − load) + K3 * delay] * [K5 / (reliability + K4)]

 

        Using the default values for K1 -> K5 this reduces down to:

 

        metric = bandwidth + delay calculated as  [(10000000/(bandwidth) + SUM(Delays)) *256]

   

        The lowest configured bandwidth is used an the delays in Microseconds / 10 are summed

        to get to the total delay.

 

        PATHS

   

        The feasible distance is the best metric along the path toward a destination network including the

        metric to the neighbor advertising that path.

        The reported distance is the metric to the destination network as advertised by an upstream

        neighbor.

        A feasible successor is a path whose reported distance is less than the feasible distance (best path)

 

        Note a feasible successor will only be designated if the successor metric is less than the reported

        distance for that route.  If no successor is in place then new queries are sent when a route goes down.

 

 

 

 

 

       

 

 

 

 

Return to Top