WHITE PAPER

The CISCO IP Routing Process

including POLICY Routing

by
Alexander Marhold
CCIE #3324, CCSI #20642, CCNP, CCDP
 route-map map-tag [permit | deny] [sequence-number] FOR POLICY ROUTING
match ip address {access-list-number | name} [...access-list-number | name]
The CISCO Routing Process
match length min max including POLICY Routing

Policy Routing
on incoming interface
set ip next-hop ip-address [...ip-address] DATA Packets
set interface type number [...type number]
selected by: Output Access-list
Input Access-list
NAT / PAT Queueing
NAT ip policy route-map map-tag Accounting

DATA no match
or deny or Recursive Lookup
set default interface type number [... type number] OUTGOING to same protocol
set ip default next-hop ip-address [...ip-address]

distribute-list {access-list-number | name} out [interface-name]

passive-interface type number
offset-list {access-list-number | name} out
INCOMING from REMOTE Routing Table offset [type number]

S ... Static
C ... Connected
offset-list {access-list-number | name} in offset [type number] x .... dynamic routing
Route-TAGs
distance weight [address mask [access-list-number | name]] Route-TAGs
distribute-list {access-list-number | name} in [type number] OUTGOING coming from other protocol
passive-interface type number (only for Link State and EIGRP)
ip access-group {access-list-number | name} in
( for selected protocol)
distribute-list {access-list-number | name} out
Metric [routing-process |autonomous-system-number]

Incoming Outgoing
ROUTE Information Routes
Route Processing Route Processing

Administrative Distance Metric

INCOMING from LOCAL OUTGOING to another protocol
0 Connected
1 Static Route redistribute protocol [process-id] {level-1 | level-1-2 | level-2} [metric
5 EIGRP Summary metric-value] [metric-type type-value] [match {internal | external 1 |
20 External BGP external 2}] [tag tag-value] [route-map map-tag] [weight weight] [subnets]
ip route prefix mask {address |
90 Internal EIGRP default-information redistribution:
interface} [distance] [tag tag]
100 IGRP default-information originate [always] [metric metric-value] [metric-type type-
[permanent]
110 OSPF value] {level-1 | level-1-2 | level-2} [route-map map-name] (RIP/OSPF)
and from connected interfaces
115 IS-IS default-information {in | out} {access-list-number | name} (IGRP/EIGRP)
120 RIP
170 External EIGRP route-map map-tag [permit | deny] [sequence-number] FOR ROUTE REDISTRIBUTION
200 Internal BGP match interface type number [...type number]
255 <don´t use> match ip route-source {access-list-number | name}[...access-list-number | name] set automatic-tag
match metric metric-value set level {level-1 | level-2 | level-1-2 | stub-area | backbone}
match route-type {local | internal | external [type-1 | type-2] | level-1 | level-2} set local-preference
match tag tag-value [...tag-value] set metric metric-value
match ip address {access-list-number | name} [...access-list-number | name] set metric-type {internal | external | type-1 | type-2}
match ip next-hop {access-list-number | name}[...access-list-number | name] set origin {igp | egp autonomous-system | incomplete}
set tag tag-value
set next-hop next-hop

 1999, PRO IN Consulting GmbH Page 2 of 18

of
! Disclaimer !
The "Cisco Routing Process" The "processes" in this paper are models
is a set of mechanisms which forward IP data for explaining the mechanisms, and are
This White Paper was done with utmost
care and thorough reviewing but is packets and which populates the IP routing table not the real implemented IOS processes.
presented "AS IS" with possible errors and by using different sources like This paper describes the above mentioned
misinterpretations.  routing updates from neighbors mechanism without focussing on particular
However none of the pictures and  connected interfaces routing protocols.
statements can be used as reference Also regarding ROUTE-MAPS this paper focuses
 static routes on IGP ( Interior Gateway Protocols) and does
regarding the behavior of the mentioned The mechanism also sends out routing updates
devices. This paper was done independent not treat the additional MATCH- and SET-clauses
eventually converting them between different which are available for BGP.
of Cisco and can never be used as routing protocols.
commitment of any party. The author and
PRO IN declares that they will not be held This paper is not based on a specific version of
Additionally "IP Policy Routing" allows IOS.
liable or responsible for any action a reader
of this White Paper will take following the to overcome the traditional destination
information given here. based routing. Topics NOT covered are:
 details of different routing protocols
All trademarks belong to their owners.  snapshot routing, ODR,…
For commanding this mechanisms a vast range
of commands and modifiers are defined in the  BGP
Cisco IOS.  route authentication
 the Link State (LS) mechanism
 QOS, COS, TOS routing
The following mechanism and behaviors  tunneling
Author: are described in detail in this white
Alexander Marhold paper: This White Paper assumes, that the reader
the general packet forwarding already has a good knowledge about IP and IP
Senior Consultant and Trainer Routing Protocols.
PRO IN Consulting GmbH process
Vienna / Austria policy routing
The structure of the paper has the picture and
[Link] routing updates and general its details always on even pages and the
behavior of routing protocols description to each picture on the page that
Copyright Notice: the INCOMING routing process and follows. Thus when printed doublesided will
its corresponding commands allow to see the picture and the explanations
1999-2001 without turning the pages.
PRO IN Training GmbH the OUTGOING routing process and
Comercial Use (Sale, Training, CBT,…) its corresponding commands The author likes to get feedback,
partly or in whole is strictly prohibited suggestions and also corrections, so please
feel free to contact him via E-mail.

 1999, PRO IN Consulting GmbH Page 3 of 18

of
 The CISCO Routing Process
including POLICY Routing

Routing Updates
Other Network
Information sources

Routing Table RIP

S .. Static
C .. Connecte d
x .. dynamic routin g
RIP

Static Routes

OSPF
Eth ernet

Connected Interfaces

 1999, PRO IN Consulting GmbH Page 4 of 18

of
Routing in General How does a router knows of its neighbor ?
How to prevent routing updates or
Again there is a difference between the routing establishing neighborship on an interface ?
Covers general topics in Routing and Routing Protocols.
Updates. - DISTANCE VECTOR protocols send out their Generally this is done using the router command
routing updates as broadcast (RIP V1, IGRP) or
Routers have 2 primary tasks: as multicasts (RIP V2) and by getting routing PASSIVE-INTERFACE <interface-name>
updates the router learns the source of these
Path Finding ( done via Routing updates. For DISTANCE VECTOR protocols this
Protocols ) command ONLY prevents the sending out of
Packet Forwarding ( Layer 3 IP function ) - LINK STATE protocols and EIGRP establish a routing updates on a particular interface.
neighborship to adjacent routers by sending However it does not prevent from getting routing
Path Finding is done by exchanging Routing HELLO-packets and control these links by updates over that interface.
information between adjacent routers. resending these HELLOs every short period.
When an ADJACENCY is found and eventually !!!HINT!!!
- In DISTANCE VECTOR routing protocols a verified the routers begin exchanging their
router forwards the networks of his routing table routing information. In order to prevent getting routing updates for
( or changes of it) to its neighbors, observing Distance Vector protocols use the router
mechanisms of SPLIT-HORIZON. Depending on !!! CAVEAT !!! command:
the protocol the network information is sent with DISTANCE 255 <netw-addr>
(subnet-)mask-information or without. In RIP LINK STATE protocols and EIGRP only uses and <wildcardmask> [ access-list ]
Version 1 und IGRP no masks are transmitted, establishes ADJACENCIES using the PRIMARY With this command al routing updates sent out
thus preventing the freedom of using IP Address of an interface. If they do not match by devices on the specified net will not be
discontigous subnets and/or VLSM (Variable the connection to the neighbor router will not be considered for entry in the routing table.
Length Subnet Masking). established.
LINK STATE protocols also verify certain For LINK STATE protocols and EIGRP
- In LINK STATE routing protocols the parameters before allowing the connection to an passive-interface prevents the establishment of
routers exchange informations regarding the ADJACENCY: adjacencies and thus the sending of any LINK
connected networks, the external routes
(interarea, static, from external routing
 same IP-subnet STATE Packets.
 equal network type However this does not prevent the router from
protocols), the connections to neighbor routers,
announcing this network as connected interface
by forwarding LSPs (Link State Packets). These  same value of timers
in its routing updates over other interfaces.
LSPs are forwarded hop-by-hop to every other the command:
OSPF treats that connected network of a
router within an area. When receiving these LSPs SHOW IP <prot> neighbor
passive-interface as STUB-NETWORK.
a router can calculate the best paths to shows the adjacencies and their status.
Also IS-IS and Integrated IS-IS have some
advertised networks.
specialities regarding the OSI or IP informations
Dependent on the routing protocol there are also
on such passive interfaces.
various DEBUG commands which show in detail
the adjacency building process.

 1999, PRO IN Consulting GmbH Page 5 of 18

of
 The CISCO Routing Process
including POLICY Routing

DATA Packets
Output Access-list
Input Access-list
NAT / PAT Queueing
NAT
Accounting

DATA
Recursive Lookup

Routing Table
S ... Static
C ... Connected
x .... dynamic routing

 1999, PRO IN Consulting GmbH Page 6 of 18

of
The Packet Forwarding What is CLASSFUL and CLASSLESS routing

Process CLASSFUL and CLASSLESS are behaviors for What is "Gateway of last Resort",
using the default route when information about a default-network, ip route [Link] [Link] ?
specific subnet is not in the routing table, but
Packets are forwarded downstream a path other subnets of that mayor network are found IP Default-Network [Link] - This
from the sender to the receiver. in the routing table. is the command that will cause a router to treat
Route information (information about the [Link] as a gateway of last resort. A
reachability of a network) is forwarded Example: router can have multiple ip default-networks
UPSTREAM from router to router. entered.
# show ip route (edited output)
This is important to consider when using … Gateway of last resort - This is the term that
blocking of routing information in order to network [Link]/16 is subnetted is applied to a routing entry in the Cisco routing
prevent access to certain networks. 2 subnets, 2 masks table that the router will use to forward packets
R [Link]/24 [120/2] [Link] eth0
R [Link]/20 [120/4] [Link] ser0
to when it lacks a more specific route. This can
The packet forwarding is done by an … be learned from a route provided by another
independent decision of each router on the path, *S [Link]/0 [0/0] [Link] ser1 router that is tagged as a default by the
using the destination address of the packet and advertising router. The ip default-network
the Routing Table as basis for finding a next-hop. The router now receives a packet on eth1 with command is one way of having a router tag a
the destination address [Link] route as a gateway of last resort.
The router will consult the routing table (or a
special forwarding table, based on the content of This address belonging to a specific subnet of IP Default-Gateway - This command is used
the routing table), comparing the destination [Link]/16 is NOT in the routing table. in routers when IP routing disabled in order to
address with the network information in the give them an address to forward packets that are
routing table and will use the most specific - With IP CLASSLESS the router will take not in their address space. Routers in boot mode
network information for a decision about the the default route and forward the packet are a good example of this situation.
outgoing path. out on Serial 1. This is done independent of
The lookup process can be recursive, that any other subnet information for that IP ROUTE [Link] [Link] establishes a default
means, that more than one lookup may be mayor network [Link]/16. route (catch-all) if no specific route is found
needed in order to find the real next-hop-address
for forwarding the packet. - When CLASSFUL routing with the !!! CAVEAT !!!
If such a next-hop or an outgoing interface is command:
NO IP CLASSLESS is selected, The [Link] route has special meaning for RIP. It is
found the router will forward the packet on the automatically installed as the local gateway of last resort. No
specified connected interface. the router would delete the packet and ip default-network [Link] is required. RIP automatically
If no route is found and also no default-route is inform the sender via ICMP that he cannot advertises the route to [Link] even if redistribute static and
available or appropriate, the router will delete forward the packet as the specified subnet a default metric are not configured.
of the mayor network [Link]/16 is not For other routing protocols the router command:
the packet and inform the sender via ICMP about DEFAULT-INFORMATION … allows specific control of
this happening. in his table. forwarding or receiving default routes

 1999, PRO IN Consulting GmbH Page 7 of 18

of
 route-map map-tag [permit | deny] [sequence-number] FOR POLICY ROUTING
match ip address {access-list-number | name} [...access-list-number | name]
The CISCO Routing Process
match length min max including POLICY Routing

Policy Routing
on incoming interface
set ip next-hop ip-address [...ip-address] DATA Packets
set interface type number [...type number]
selected by: Output Access-list
Input Access-list
NAT / PAT Queueing
NAT ip policy route-map map-tag Accounting

DATA no match
or deny or Recursive Lookup
set default interface type number [... type number]
set ip default next-hop ip-address [...ip-address]

Routing Table
S .. Static
C.. Connected
x .... dynamic routing

 1999, PRO IN Consulting GmbH Page 8 of 18

of
IP Policy Routing !!! CAVEAT !!! interface Serial3/0.31 multipoint
description INTERNET ACCESS
If there is an outgoing interface defined in a ip address [Link] 255.255…
IP Policy Routing overcomes the normal SET-clause, this interface must be up and be of a ip policy route-map OUT-to-PIX
destination based routing paradigma by allowing point-to-point type. frame-relay map ip [Link] 501
different criteria as basis for a routing decision. !
Among those criterias are: If there is a next-hop-address specified in the interface FastEthernet4/1.24
SET-clause this address have to be a real next- description PIX-OUT
 the incoming interface hop-address. That means that it must be an encapsulation isl 24
 selection by extended access-lists address of a device belonging to a directly ip address [Link] [Link]
 precedence levels connected network. (The Router will not do a ip policy route-map PIX-to-OUT
 packet sizes recursive lookup for the next-hop-address) !
 … route-map PIX-to-OUT permit 10
If the above mentioned requirements are not match ip address 1
But still one paradigma stays valid: met, the router will use the normal Routing table set ip default next-hop [Link]
"The router only makes a local decision about based route decisions and ignore the SET !
the next hop, i.e. where to send the packet out" parameters. route-map OUT-to-PIX permit 10
To overcome this one you need either Tunneling match ip address 1
or MPLS (Multiprotocol Label Switching). Example: set ip default next-hop [Link]
The same Frame Relay interface is used as !
IP Policy Routing uses ROUTE-MAPS for connection to the outside world AND as access-list 1 permit any
defining the matching packets and for setting connections to Remote offices. The Firewall is !
actions. placed in to VLANs on a Fast Ethernet attached PIX OUT
ROUTE-MAPS define a numbered sequence of switch
MATCH and SET clauses , where the SET
defines the actions to be done for packets Fast Ethernet Interface
Internet
matching the MATCH clauses.
IP POLICY ROUTING is applied to incoming
packets on interfaces by using the Interface

FRAME RELAY Interface

command:
OUT-to-PIX

Internet [Link].
FR
IP POLICY ROUTE-MAP route-map-name ISL
PIX-to-OUT

In case of no match found or when there is no

SETclause specifiying a next-hop or an outgoing Routing
interface, then after the ROUTE-MAP the normal Table
routing table is used to find a next-hop-
Remote Offices
address or outgoing interface.

 1999, PRO IN Consulting GmbH Page 9 of 18

of
 The CISCO Routing Process
including POLICY Routing

Routing Table
S.. Sta tic
C ... Connecte d
x .. dynamic routin g
Route-TAGs
Route-TAGs

Metric
Incoming Outgoing
ROUTE In fo rmation Route s
Route Processing Route Processing

Administrative Distance Metric

 1999, PRO IN Consulting GmbH Page 10 of 18

of
  Sometimes I am wondering why the developers
Routing Information gave the second best distance of 1 to static routes
This allows Failover of routes also without dynamic
routing protocols, when for example 2 static routes
Processes: entered by an administrator, as so called "Quick Fixes"
by using static routes are often the cause of
with different Administrative Distances for 2 outgoing
interfaces or 2 different next-hop-addresses are
General Considerations reachability and routing-loop problems.  defined.

§4 Route REDISTRIBUTION is only used for §8 Routing processes are relying on a

A lot of problems and confusion arises from the fact
outgoing routing updates. consistent metric, in order that every router
that some basic principles in the Routing information
find the best path in a way, that all routes are
process are not correctly understood.
In Principle: Route redistribution means that routes of leading in the same direction.
Therefore in this chapter I will give some fundamental one routing protocol in the routing table will be sent
laws and principles and describe their consequences: out, converted to another routing protocol on As the base of metrics is different for different routing
interfaces configured for routing updates of that protocols, a direct conversion of metrics from one
second routing protocol. routing protocol to another is generally not possible.
§1 The mechanism of processing incoming
When having more than one routing process default
routing updates is COMPLETELY separated from
§5 Static Routes defined with a next-hop- metric information has to be used. This default hides
the mechanism of creating outgoing routing
address are considered one hop away and have the correct information about the best path and this
updates.
a default Administrative Distance of 1 inconsistency will lead to not optimal routing and also
often to ROUTING LOOPS in MUTUAL
The fact that a route is found in the routing table of a REDISTRIBUTION (i.e. more than one redistribution
router is a prerequisite but NOT necessarily sufficient If the next-hop-address specified in the static route is
not a REAL next-hop-address (i.e. not an address in a points).
criteria for an outgoing routing update.
directly connected network) the router will do
recursive lookups to find this REAL next-hop-address. §9 Routing is a STATEFUL process. Depending
§2 The original routes of every configured on the current information in the Routing Table
routing process of a router will be considered, different actions can happen, even when the
when decisions about which will enter the §6 Static Routes defined with an outgoing
interface are treated like connected networks same routing information is received.
routing table will be made.
( i.e. networks that are 0 hops away) and thus
having a default Administrative Distance of 0.  There are examples, where a routing was correct,
This means that route REDISTRIBUTION is never used but after the shutdown and restart of an interface the
in the incoming route processing. correct state was never reached again.[See page 15]
Therefore static routes defining an outgoing interface
should be used only when the destination is on that
§3 If more than one information of a route is
connected network.
found, the incoming route process will use first
USAGE: When the connected network is address
the ADMINISTRATIVE DISTANCE and then the
translated, you need a static route for the outside
METRIC for deciding which route will be
network pointing to that inside hidden network.
established in the routing table.
§7 Static Routes where the outgoing interface is
Cisco IOS has a predefined Administrative Distance for
down or the next-hop-address is not reachable
each Routing Protocol which allows to prefer more
are removed from the routing table unless the
trusted information sources over less trusted one.
parameter PERMANENT is specified.

 1999, PRO IN Consulting GmbH Page 11 of 18

of
 The CISCO Routing Process
including POLICY Routing

INCOMING from REMOTE Routing Table

S.. Static
C ... Connecte d
x .. dynamic routing
offs et- lis t {access-list- number | name} in offs et [type number] Route-TAGs
distance weig ht [a ddress mask [access-lis t-number | name]]
distrib ute -lis t {access-list-number | name} in [type number]
passive-in terface type number (only for Lin k Sta te and EIGRP)
ip access-group {access-lis t- number | name} in
( fo r sele cted proto col)

Metric
Incoming
ROUTE In fo rmatio n
Route Processing

Administrative Distance
INCOMING from LOCAL
0 Connecte d
1 Sta tic Route
5 EIG RP Summary
ip route prefix mask {address | 20 Exte rnal BGP
inte rfa ce} [d is ta nce] [ta g ta g] 90 Inte rnal EIG RP
[permanent] 100 IGRP
and fr om connected in te rfa ces 110 OSPF
115 IS-IS
120 RIP
170 Exte rnal EIG RP
200 Inte rnal BGP
255 <don´t use>

 1999, PRO IN Consulting GmbH Page 12 of 18

of
The INCOMING Routing Monitoring the INCOMING Route process Useful commands for changing behaviour
of the Incoming Roue process
Process All the input and results of this incoming routing
decision process can be monitored with the These commands can have different goals:
command:
The incoming Routing process is responsible for DEBUG IP ROUTING  prevent routing information from
populating the Routing table. Unfortunately the debug output is somewhat entering the routing table
cryptic and therefore not easy to read. distribute-list xxx in … (not for Link State)
At startup this process enters the static and Here an example with the output of a RIP distance 255 …
connected networks for all interfaces which are routing change and its real meaning. passive-interface …(for Link State & EIGRP)
UP and then for each route received via any  change the priority of some information
Routing potocol this process checks if this is a RouterA# debug ip routing sources or for some commands
better route (considering Adminstrative Distance RT: flushed route to [Link] via [Link]  by changing the ADMINISTRATIVE DISTANCE
and metric) than another instance of the same (Serial0) distance <0…154> …
route already in the table. RT: no routes to [Link], entering holddown
If a better route is found, this one is installed  invalid timer expired no routes to  by changing the METRIC
and the other one is removed from the table. [Link], therefore entering holddown offset-list xxx in … (not for Link State)
The different routing processes also inform the RT: flushed route to [Link] via [Link]  manually adding additional routing
incoming routing process about any routes for (Serial0) information
which regular routing updates are missing, or  advertising [Link] via [Link] ip route …
which route to remove. (Serial0) as unreachable
In order to overcome incorrect routing In order to prevent possible routing loops when
information Distance Vector routing processes "show ip route" shows us getting redistributed (external routes) EIGRP
also sets routes into a temporary holddown … uses the higher administrative distance of 170
before reconsidering new routing information or R [Link]/24 is possibly down, instead of the default of 90.
before deleting this route. routing via [Link], Serial1
LINK STATE processes directly remove or In OSPF you also can use TAGs for marking routes
…
replace routes after running the SPF-calculation. and then applying actions to tagged routes.
EIGRP when a feasability successor is found will RT: garbage collecting entry for [Link] NOTE: For Link State protocols you cannot apply incoming
enter the new information direct into the routing  flush timer expired terminating holddown for filters as those protocols transfer not routes but LINK STATE
table, or will set the route to a state of ACTIVE [Link] Packets.
and ask the neighbor(s) for a new route to the after that the next update info for this network
destination. will be used In BGP you can specify a route-map which can
RT: add [Link]/24 via [Link], rip modify parameters like metric and tag when BGP
RIP V1 and IGRP will never establish an metric [120/2] sends routing information to the local routing
incoming mayor route, when they have a local table:
subnet-route of that network in their routing table-map route-map name
table.

 1999, PRO IN Consulting GmbH Page 13 of 18

of
 The CISCO Routing Process
including POLICY Routing

OUTGOING to same protocol

distr ib ute -lis t {access-lis t- number | name} out [in te rfa ce-name]
passiv e-in te rface type number
offs et-lis t {access-lis t- number | name} out
Routing Table offs et [ty pe number]

S .. Static
C .. Connecte d
x .. dynamic routin g

Route-TAGs
OUTGOING coming from other protocol

dis trib ute -lis t {access-lis t-number | name} out

[r outin g-process |a uto nomous-system-number]

Outgoing
Route s
Route Processing

Metric
OUTGOING to another protocol
redis tr ib ute proto col [p rocess-id] {le vel- 1 | le vel- 1-2 | level-2} [metr ic
metr ic -value] [metr ic-ty pe ty pe-valu e] [m atc h {in te rnal | exte rnal 1 |
exte rnal 2}] [ta g ta g-value] [r oute -map map-ta g] [w eig ht weight] [s ubnets ]
defa ult- in fo rmatio n redistr ib utio n:
defa ult- in fo rmatio n orig in ate [a lw ays] [metric metr ic -valu e] [metr ic -ty pe ty pe-
valu e] {le vel- 1 | le vel- 1-2 | le vel-2} [r oute -map map-name] (RIP/O SPF)
defa ult- in fo rmatio n {in | out} {access-lis t- number | name} (IG RP/EIG RP)

route -map map-ta g [p ermit | deny] [s equence-number] FOR ROUTE REDISTRIBUTIO N

matc h in te rfa ce ty pe number [. .ty pe number]
matc h ip route -source {access-lis t-number | name}[.. a ccess-lis t- number | name] set auto matic -ta g
matc h metr ic metr ic-valu e set le vel {level- 1 | le vel-2 | le vel-1-2 | stu b-area | backbone}
matc h route -type {lo cal | in ternal | exte rnal [ty pe-1 | ty pe-2] | le vel- 1 | le vel- 2} set lo cal- prefe rence
matc h ta g ta g-valu e [. .ta g-valu e] set metr ic metr ic -valu e
matc h ip address {access-lis t-number | name} [. .a ccess-lis t- number | name] set metr ic -ty pe {in te rnal | exte rnal | ty pe-1 | ty pe-2}
matc h ip next-hop {access-lis t- number | name}[. .a ccess-lis t- number | name] set orig in {ig p | egp auto nomous-syste m | in complete }
set ta g ta g-value
set next- hop next- hop

 1999, PRO IN Consulting GmbH Page 14 of 18

of
The OUTGOING Routing For RIP V1 and IGRP the following also is
considered:
Example for a Redistribution problem
which is state-dependent:
Update Process A subnet route of a mayor network is converted to the
(summary) mayor route when it is sent out on
Configuration of R4 and R5:
interfaces that do not belong to a (sub)net of that
mayor route. (Discontigous Subnet Rule)
router rip
The outgoing Routing update process is network x.x.x.x
responsible for informing the neighboring routers redistribute eigrp 1000
For OSPF there are some special rules, where
about its network information. default-metric 1
Area Border Routers (ABR) can inject default
For Distance Vector Routing protocols this is passive-interface Serial 1
routes into stub areas. Also DISTRIBUTE-LIST
the local information about the best routes and router eigrp 1000
OUT can only be applied to external routes and
their metric (hence the content of the Routing network y.y.y.y
you cannot specify an interfacename in OSPF.
Table). redistribute rip
For Link State Routing protocols this is the default-metric 1000 100 250 100 1500
information about the local networks, external Generally by using the shown commands you can
follow 4 different tasks: passive-interface Serial 0
routes and the neighbors via LSPs. R3 normally learns about the network [Link]/8
 make networks invisible by blocking the via a routing update from R2 with 2 hops and
Outgoing routing updates for a certain routing
forwarding of routing information forwards this information to R5 with 3 hops
protocol are only sent when the following
 redistribute (forward and translate) routing R5 gets information about [Link]/8 via R4 with
conditions are all met:
information from one protocol to another the metric [170/10245] ( it is an external EIGRP
 the network is in the routing table.  change the metric to force the others to route) and via R3 with [120/3] as RIP-route.
 the network is either specified via the NETWORK prefer specific paths So R5 will establish the RIP-route and use R3 as
command or coming from another protocol via a  summarize routing information to next hop. Obeying the SPLIT HORIZON rule it
REDISTRIBUTION command
decrease the amount of routes and to increase will never send out the information back to R3.
 obeys the SPLIT-HORIZON rule: the network was
not learned from the same interface ( or is not the stability
When the connection between R2 and R3 breaks,
identical to the connected network) R3 will not send information about [Link]/8. R5
 the network is not excluded from update via
[Link]/8 will now use the routing information derived via
applied access-lists or route-maps using the
DISTRIBUTE command. external EIGRP from R5 and forward this
0 hops
 the outgoing interface is not specified as PASSIVE best route to10.0.0.0/8 information as redistributed information via Rip
before shutdown of R2-R3 R1 RIP to R3. R3 gets now the information about
 if the network is a specified summary, at least a 0
subnet of that summary route is in the routing [Link]/8 with the metric [120/1] and next hop
best route to10.0.0.0/8 1 hop
table.
after shutdown of R2-R3
R5 into its routing table. When the link R2-R3
 forwarding of default information is implicitly R10 RIP comes up again, the information from R2 about
(RIP) or explicitly allowed via the 3 hops 1 [Link]/8 with [120/2] will not be used and R3
DEFAULT-INFORMATION … command. EIGRP
2 hops will continue to use the way via R5 to reach that
 for OSPF: sending of LSPs to that neighbor is not 2
network.
prohibited. R4
RIP
R2
2

before
shutdown 3 hops
3
RIP
R5 R3
 1999, PRO IN Consulting GmbH Page 15 of 18
1
of 1 hop after shutdown of
R2-R3
 route-map map-tag [permit | deny] [sequence-number] FOR POLICY ROUTING
match ip address {access-list-number | name} [...access-list-number | name]
The CISCO Routing Process
match length min max including POLICY Routing

Policy Routing
on incoming interface
set ip next-hop ip-address [...ip-address] DATA Packets
set interface type number [...type number]
selected by: Output Access-list
Input Access-list
NAT / PAT Queueing
NAT ip policy route-map map-tag Accounting

DATA no match
or deny or Recursive Lookup
set default interface type number [... type number] OUTGOING to same protocol
set ip default next-hop ip-address [...ip-address]

distribute-list {access-list-number | name} out [interface-name]

passive-interface type number
offset-list {access-list-number | name} out
INCOMING from REMOTE Routing Table offset [type number]

S ... Static
C ... Connected
offset-list {access-list-number | name} in offset [type number] x . .. dynamic routing
Route-TAGs
distance weight [address mask [access-list-number | name]] Route-TAGs
distribute-list {access-list-number | name} in [type number] OUTGOING coming from other protocol
passive-interface type number (only for Link State and EIGRP)
ip access-group {access-list-number | name} in
( for selected protocol)
distribute-list {access-list-number | name} out
Metric [routing-process |autonomous-system-number]

Incoming Outgoing
ROUTE Information Routes
Route Processing Route Processing

Administrative Distance Metric

INCOMING from LOCAL OUTGOING to another protocol
0 Connected
1 Static Route redistribute protocol [process-id] {level-1 | level-1-2 | level-2} [metric
5 EIGRP Summary metric-value] [metric-type type-value] [match {internal | external 1 |
20 External BGP external 2}] [tag tag-value] [route-map map-tag] [weight weight] [subnets]
ip route prefix mask {address |
90 Internal EIGRP default-information redistribution:
interface} [distance] [tag tag]
100 IGRP default-information originate [always] [metric metric-value] [metric-type type-
[permanent]
110 OSPF value] {level-1 | level-1-2 | level-2} [route-map map-name] (RIP/OSPF)
and from connected interfaces
115 IS-IS default-information {in | out} {access-list-number | name} (IGRP/EIGRP)
120 RIP
170 External EIGRP route-map map-tag [permit | deny] [sequence-number] FOR ROUTE REDISTRIBUTION
200 Internal BGP match interface type number [...type number]
255 <don´t use> match ip route-source {access-list-number | name}[...access-list-number | name] set automatic-tag
match metric metric-value set level {level-1 | level-2 | level-1-2 | stub-area | backbone}
match route-type {local | internal | external [type-1 | type-2] | level-1 | level-2} set local-preference
match tag tag-value [...tag-value] set metric metric-value
match ip address {access-list-number | name} [...access-list-number | name] set metric-type {internal | external | type-1 | type-2}
match ip next-hop {access-list-number | name}[...access-list-number | name] set origin {igp | egp autonomous-system | incomplete}
set tag tag-value
set next-hop next-hop

 1999, PRO IN Consulting GmbH Page 16 of 18

of
Summary: The BIG picture  Routing Protocols rely on a consistent
metric
For further information

 REDISTRIBUTION of routes means a loss Recommended Requests for proposals (RFCs):

The CISCO IP Routing Process and its of topology information RFC1812 Requirements for IP Version 4 Routers.
mechanisms are quite complicated. But a  Routing is a STATEFUL process, where F. Baker. June 1995.(Status: PROPOSED
thorough understanding is necessary to the incoming routing information is STANDARD)
troubleshoot or even better to avoid problems. considered in relation to the current  general information about Routing:
routing table information. RFC1771 A Border Gateway Protocol 4 (BGP-4).
The basic points are:  The fact that a route is in the routing [Link] & T. Li. March 1995.
table does not necessarily mean that the (Status: DRAFT STANDARD)
 Routing is done hop-by-hop, each router route is also used in outgoing routing RFC2328 OSPF Version 2. J. Moy.
independently decides on which interface updates April 1998. (Status: STANDARD)
to forward a packet.  the adjacency process for Link State and RFC2453 RIP Version 2. G. Malkin.
 The router treats incoming and outgoing EIGRP as basis for exchanging updates November 1998. (Status: STANDARD)
routing mechanisms as completely between routers
separate processes.  the different behavior of Routing Recommended Books:
 Decisions about which route to add or protocols regarding summarization and CCIE Professional Development: Routing TCP/IP
remove from the routing table are based VLSM Volume 1, [Link] ISBN: 1-57870-041-8
on  excellent description of Routing
ADMINISTRATIVE DISTANCE and  best description of the mechanisms of EIGRP
METRIC What is CISCO-specific in that area ?  very good treatment of all Routing Protocols
 Routes of all configured routing CCIE Professional Development: Large-Scale IP
processes are considered for the routing  The use of ADMINISTRATIVE DISTANCE as Network Solutions, K. Raza, S. Asad, M. Turner
table first considered parameter for incoming route ISBN: 1-57870-084-1
 REDISTRIBUTION is only used when decisions.
 good examples of routing design
considering outgoing routing updates  IGRP and EIGRP are Cisco-developed and
 excellent examples of redistribution
 POLICY ROUTING allows to overcome proprietary protocols.
 good description of Routing Protocols
the normal destination based routing  REDISTRIBUTION, metric handling on
Internet Routing Architectures, [Link]
 Policy Routing is applied on packets redistribution is not covered in standards.
ISBN:
incoming on specified interfaces  the treatment and forwarding of DEFAULT-
ROUTES is not covered in standards.  best book on BGP
 ROUTE-MAPS are a mechanism for using
 an extensive set of DEBUG commands for OSPF [Link]
additional parameters for selection and
monitoring the router behavior. ISBN:
also a mechanism for setting or changing
different parameters  OSPF explained by the developer of that
 ROUTE-MAPS are used for POLICY protocol
ROUTING and for a controlled WWW-locations:
REDISTRIBUTION of Routing Updates [Link]
[Link]
[Link]

 1999, PRO IN Consulting GmbH Page 17 of 18

of
 PROFESSIONAL INFORMATION
NETWORKS

PRO IN is a paneuropean company focussing on

3 areas:
 Training
 Consulting
 Professional Services

With offices in Austria, Germany and Spain we

offer CISCO authorized trainings at the highest
possible level. Thus PRO IN is honored as
"Distinguished Trainings Partner" by CISCO