How to Select a Best Routing Protocol for your Network

woonsocketpoliticalΔίκτυα και Επικοινωνίες

28 Οκτ 2013 (πριν από 3 χρόνια και 8 μήνες)

57 εμφανίσεις

Canadian Journal on Network and Information Security Vol. 1, No. 6, August 2010


Routing protocols play a vital role in the performance and
reliability of IP networks. Two types of routing are used to
determine the best route between the routers. Static routing is
configured manually by the administrator and dynamic routing
is configured by routing protocols such as RIP (Routing
information Protocol), EIGRP (Enhanced Interior Gateway
Routing Protocol), OSPF (Open shortest Path First), IS-IS
(Intermediate system-intermediate system) and BGP (border
gateway protocol). BGP is exterior gateway protocol used
between different autonomous systems. IGRP & EIGRP are
CISCO proprietary routing protocols that can only be
configured on CISCO routers whereas RIP, OSPF and IS-IS
can be used on different vendor routers. This paper finally
concludes that OSPF is the best option for the large networks
because it supports VLSM (variable length subnet mask),
CIDR (classless inter-domain routing), provide loop free
routing, converge quickly, scalable, efficient use of
bandwidth, small update packets and it can be used between
different vendor routers and BGP will be used between
different autonomous systems.
Index Terms— variable length subnet mask, classless inter-
domain routing, Open shortest Path First, Intermediate system
- intermediate system, Enhanced Interior Gateway Routing
Protocol, Routing information Protocol.
1.0 Introduction
The routing of packets between IP networks is carried out by
two different ways i.e. static routing & dynamic routing. In
static routing “Remote networks are added to the routing
table” [3] manually by the administrator. Static routing cannot
be used for larger networks because if any change occurs in
topology then the network administrator will have to
reconfigure the whole network manually and it is very time
consuming. Dynamic routing is widely used for big IP
networks. The protocols such as RIPv1 & IGRP are classfull
routing protocols where as RIPv2, EIGRP, OSPF, IS-IS &
BGP are classless routing protocols. Classless routing
protocols include the subnet mask with the network address in
their routing updates & are necessary when the mask cannot
be assumed or determined by the value of the first octet
whereas classfull routing protocols don’t include the subnet
mask in the routing update. Classless routing protocols also
support VLSM (variable length subnet mask) & CIDR
(classless inter-domain routing). “CIDR uses VLSM to
allocate IP addresses to subnets according to individual need
rather than by class. This type of allocation allows the
network/host boundary to occur at any bit in the address.
Network can be further divided or subnetted into smaller and

smaller subnets”[3]. EIGRP, IS-IS, OSPF etc. are used within
an autonomous system (AS) where as BGP is used to share
routing information between the different autonomous
systems. “An AS is a group of routers that share similar
routing policies and operate within a single administrative
domain”[3]. Dynamic routing protocols can be further
classified into two classes i.e. distance vector routing
protocols and link state routing protocols. RIP and IGRP both
distance vector routing protocols use hop count as metric. Like
RIP V1, IGRP does not support VLSM & CIDR [3]. EIGRP is
hybrid distance vector routing protocol commonly used in
large networks. It supports VLSM, CIDR, fast convergence,
partial updates and provides loop free network but it is not
internet standard due to vendor limitation. IS-IS & OSPF both
are link state routing protocols. OSPF is mostly used now a
day for large networks because it is open standard and
provides many advantages over distance vector. IS-IS is
almost similar to OSPF but it doesn’t support route tag,
demand circuit and three types of stubby areas [4,5,6]. This
paper will analyze and compare the main features of all
routing protocols.
2.0 Feature Comparison of Routing Protocols
Two types of dynamic routing protocols are used in IP
networks to share routing information.
1. IGP (Interior gateway protocol)
2. EGP (Exterior gateway protocol)

IGP protocols such as EIGRP, IS-IS and OSPF are used in an
autonomous system managed by one administrative domain or
a group of administrator with the same routing policy where as
Protocols that run outside an enterprise, or between
autonomous systems (ASs), are called exterior gateway
protocols (EGPs). Typically, EGPs are used to exchange
routing information between Internet Service Providers (ISPs).
EGP such as BGP can also be used between a customer's AS
and the provider's network. [3]
2.1 Classfull/Classless Routing Protocols (VLSM, CIDR
and supernetting)
Classful routing protocols such as RIPv1 and IGRP don’t send
the subnet mask within its routing updates. That makes the
routing table big and also increases the load on the link. These
protocols also don’t support VLSM, CIDR and super-netting.
CIDR provides the following advantages.
• Efficient use of IPv4 addresses space.
• Prefix aggregation, which reduced the size of routing
How to Select a Best Routing Protocol for your Network

Muhammad Tayyab Ashraf
Canadian Journal on Network and Information Security Vol. 1, No. 6, August 2010


Classless routing protocols i.e. RIPv2, EIGRP, OSPF, BGP
and IS-IS are classless routing protocols. All these support
VLSM, CIDR and super-netting. They summarize the route
and make the routing table small. [7, 8]
In cooperate networks, Classless routing protocols are used
because these protocols support VLSM and CIDR that
summarize the routes and make the routing table small. It also
improves the efficiency of the router and also utilizes the link
more efficiently.
2.2 Routing Protocols Metric
Metric is used by the routing protocols to determine that
which route should be inserted into the routing table if there
are more than one route exist for the same destination
network. Routing protocols use different metrics i.e. RIP uses
hop count as metric. Hop count is the number of routers
between source and destination. IGRP & EIGRP routing
protocols use bandwidth, delay, reliability and load as metric.
The calculation is accurate but it takes more memory and
processing time of the router. Both the protocols run on the
CISCO routers. IS-IS metric is cost and OSPF routing
protocol uses cost (higher bandwidth indicates lowest cost) as
metric. This property of OSPF makes it more efficient than
other protocols and it can run on different vendor routers. BGP
does not use technical metrics. Instead, BGP makes routing
decisions based on network policies, or rules. [3].

Routing Protocol Metric
RIPv1, RIPv2 Hop count

Bandwidth, Load, Delay,
Reliability, MTU (EIGRP factor
of 256)
IS-IS Cost (6 bit metric value
between 1 to 63)
OSPF Cost (Higher bandwidth
indicates lowest cost)
BGP Network policies or rules

Table 1: Metric Comparison

2.3 Convergence & Routing Loops
When a network topology or any change occurs in the network
then all the routers belonging to that network update their
routing tables. Routing loops create problems for the networks
to be converged. Fast convergence and loop free networks are
designed for cooperate sector. Routing protocols use different
methods for fast convergence and for the elimination of
routing loops. RIP protocol is prone to routing loops and its
convergence is very slow. Both RIP & IGRP use Bellman-
Ford distance vector algorithm for convergence. Bellman-
Ford algorithm “makes each router periodically broadcast its
routing tables to all its neighbors. Then a router knowing its
neighbors' tables can decide to which destination neighbor to
forward a packet”. EIGRP is free from routing loops and it
uses DUAL (diffusing update algorithm). DUAL uses the
concept of Successor, Feasible Distance (FD), Feasible
Successor (FS), Reported Distance (RD) or Advertised
Distance (AD), Feasible Condition or Feasibility Condition
(FC) and it provides loop free paths, loop free backup paths
which can be used immediately, minimum bandwidth usage
and provides the fast convergence. OSPF & IS-IS are also free
from routing loops and both run the Dijkstra algorithm to
calculate the shortest-path first (SPF). Dijkstra algorithm
calculates the cost of the links to calculate the route. The cost
of the link is actually the bandwidth of the link. Higher
bandwidth indicates lower cost. EIGRP takes less than 3
seconds to be converged where as OSPF converges within 10
seconds [9, 10]. BGP constructs a graph of autonomous
systems based on the information exchanged between BGP
neighbors. BGP views the whole internetwork as a graph, or
tree of autonomous systems and make the network loop free.
Routing Protocol Prone to Routing Loop
RIPv1 Yes
RIPv2 Yes


Table 2: Routing loops
2.4 Scalability
RIPv1 and RIPv2 only supports up to 15 hop counts so these
cannot be scalable. Similarly IGRP & EIGRP are scalable up
to 255 and 224 hop counts respectively. Both the CISCO
protocols use the autonomous number concept in which all the
routers remains in a single administrative domain. IS-IS
routing protocol support large networks up to 1024 routers and
it is also scalable. IS-IS uses different level of routing Level1
and Level2. OSPF is the only routing protocol that supports
very large networks with unlimited routers (hop count). But
the best option is to use up to 50 routers per area. In OSPF a
large network is divided into small multiple areas. This
functionality of OSPF makes it scalable and efficient. More
areas can be defined in the existing network and can be
managed easily [5, 6]. BGP is also scalable and it supports
multi-area routing.
2.5 Full/Partial Routing Updates
Full routing updates consume bandwidth that makes the link
slow as well as the router needs more memory to process these
updates. In order to improve the efficiency of the link and for
fast convergence, partial or bounded updates are used.
Distance vector routing protocols such as RIPv1 and IGRP
broadcast full routing tables to its neighbors at regular
intervals that make the link slow and consume more
bandwidth. EIGRP and link state protocols such as OSPF and
IS-IS don’t send complete routing tables to its neighbors. Both
EIGRP and OSPF routing protocols are event trigged. BGP
also sends the partial routing updates and it sends different
types of messages. [3, 11, 12]
View of Topology
Two types of topologies are used by the routing protocols to
view the network.
• Flat Topology
• Hierarchical Topology
Canadian Journal on Network and Information Security Vol. 1, No. 6, August 2010


In flat topology, the routing protocol sends and receives the
full routing updates from its neighbors and it doesn’t have
common view of network whereas in hierarchical topology,
the routing protocol has a common view of the entire network
and it only sends partial or bound updates to specific
interface/network. The protocols that support hierarchical
topology are mostly used now a days in cooperate networks.
RIP and IGRP uses flat topology. These protocols broadcast
the complete routing updates to all the interfaces whereas the
EIGRP, OSPF and IS-IS use the hierarchical design topology.
OSPF & IS-IS uses areas to form hierarchical design
topologies. Every router running under these routing
protocols has a best route to reach the destination within the
network. This functionality of EIGRP and link state protocols
reduce the overhead by multicasting only partial/incremental
updates to specific interfaces and it also improves the
performance of the network. BGP constructs a graph or tree of
autonomous systems and view the entire internetwork. [3,7,8].

Protocol Topology Overhead Performance Common
view of
RIPv1 Flat

Increases decreases No
RIPv2 Flat

Increases decreases No

Increases decreases No
EIGRP Hierarchical

Reduced Increase yes
OSPF Hierarchical

Reduced Increase yes
IS-IS Hierarchical

Reduced Increase Yes
BGP Hierarchical
(Graph, tree)
Reduced increase yes

Table 3: Performance and Overhead Comparison

2.6 Neighbor, Topology & Routing Table
EIGRP & OSPF both routing protocols maintain the neighbor,
topology and routing tables to make the hierarchical view of
the entire network. Neighbor table contains the information of
the directly attached neighbors by sending regular hello
updates. Topology table/databases are used to contain the
information of all links status between the routers. Then both
the routing protocols use these two tables/databases to
calculate the best route and maintain it in the routing table.
EIGRP uses the concept of successor and feasible successor
where the first one is primary route to destination calculated
by DUAL algorithm and feasible successor is the backup route
for the same destination and it is kept in topology table. By
using these tables, EIGRP converged very fast within 3
seconds as compared to other routing protocols. OSPF also
maintain these 3 tables but it uses the concept of DR
(designated router) and BDR (backup designated router). DR
is the central focal point. Any change in the network is
reported to DR which then update to all the routers. If the DR
goes down then BDR take over the charge of DR. OSPF also
converged fast within 10 seconds. IS-IS also makes
adjacencies with neighbors and uses the same concept of DIS
(designated intermediate system) as central point. But it don’t
use backup router. In case of DIS failure or any new router
with higher priority entered in the network, the new election
takes place. RIP and IGRP routing protocols are distance
vector routing protocols and these protocols don’t use these
tables [11, 13, 14].
3.0 Future Improvement

Most commonly used routing protocols are EIGRP, OSPF. IS-
IS is also getting popularity. These protocols can be improved
in future to make more efficient.
OSPF is the most efficient open standard protocol currently
used in large networks. Mostly IP protocol is used in the world
for larger networks. OSPF doesn’t support multi protocols like
IPX and AppleTalk. It can be improved by adding the
multiprotocol support.
EIGRP is the CISCO proprietary routing protocol that only
runs on CISCO routers. It is the best routing protocol for the
networks having all CISCO routers. It can be improved in
future by implementing different vendor support functionality.
IS-IS is also a very good routing protocol. Currently it is not
used widely. It is similar to OSPF and use the DIS (designated
intermediate system) as a central point. But it doesn’t have
backup system. This routing protocol can be improved in
future by adding the functionality of BDIS (backup designated
intermediate system), support for 3 types of stubby areas,
demand circuits and the support for NBMA networks [19].

4.0 Conclusion

OSPF routing protocol is the best option for all sized networks
as compared to other interior routing protocols because it is
easily scalable. OSPF is a classless routing protocol and it
supports VLSM, CIDR and supper-netting. It divides the
larger network into small areas that can be easily manageable.
OSPF uses the hierarchical view of topology that makes it
more efficient and also increases the performance. EIGRP is
CISCO proprietary and cannot run on different vendor routers.
IS-IS provides almost the same functionality as OSPF. But in
OSPF if the DR fails then the BDR takes the position of DR
without any delay but in IS-IS if the DIS goes down then
again the election takes place on priority bases. There is not a
concept of backup router. So it takes some time for the
election to take place. IS-IS is not popular at this stage and its
configuration is difficult. OSPF is widely used in all networks.
BGP (exterior gateway protocol) will be used between
different autonomous systems. Static routing cannot be used
for larger networks because it is difficult to configure and if
topology changes then the administrator have to reconfigure
the whole network manually. [3,5,18,19].
• OSPF for Interior Routing (within autonomous
• BGP for exterior Routing (between autonomous

Canadian Journal on Network and Information Security Vol. 1, No. 6, August 2010


Protocol OSPF
Scalability Divide the network into many small
areas. New areas can be added.
Convergence Send trigger updates when change occur
and converged within 10 sec.
Classless Support VLSM, CIDR, Supper-netting
and route aggregation.
Reliability Store multiple paths for a single
Availability Support load balancing.
Bandwidth Efficient use of bandwidth by sending
hello and partial updates.
Not prone to routing loops. Use Dijkstra
Security Use MD5 authentication and authenticate
the source of hello packets.
Metric Cost (Bandwidth)
Hop count Unlimited but preferable 50 routers per
Hierarchical view of Topology (maintain
topological database for all links).
Support Broadcast multi-access networks,
non broadcast multi-access networks
(NBMA) and point to point networks.
Route Tag Tag the routes with external or internal.

Table 4: Detail Comparison between Protocols


[1] Faraz Shamim, Zaheer Aziz, Troubleshooting IP
Routing Protocols, Cisco Press, 2002, pp. 11
[2] Ramadas Shanmugam, Using TCP/IP, Que
Publishing 2002, pp. 262
[3] accessed on 20/10/2008
[4] Wendell Odom, CCNA ICND2, Cisco Press, 2008,
pp. 314-315
[5] John T. Moy, OSPF: Anatomy of an Internet Routing
Protocol, Addison-Wesley, 1998, pp.101
[6] Hannes Gredler, The Complete IS-IS Routing
Protocol, Springer, 2005, pp. 79-82
[7] Scott M. Ballew, Managing IP Networks with Cisco
Routers, O'Reilly, 1997, pp.132
[8] Wendell Odom, CCNA ICND2, Cisco Press, 2008,
pp. 314-315
[9] Todd Lammle, CCNA Study Guide, Sybex, 2005, pp.
[10] Wendell Odom, CCNA INTRO, Cisco Press, 2004,
pp. 411-421
[11] accessed on 30/10/2008
[12] Todd Lammle, CCNA Study Guide, Sybex, 2005, pp.
[13] W Richard Stevens, Gary R Wright, TCP/IP,
Addison-Wesley, 1994,
[14] W. Odom, Computer Networking First-Step, Cisco
Press, 2004, pp. 154, 580
[15] Uyless D. Black, IP Routing Protocols, Prentice Hall,
.Shtml accessed on 01/11/2008
accessed on 03/11/2008
[18] accessed on
sld004.htm accessed on 05/11/2008
BSCN.php accessed on 06/11/2008
VS.-OSPF.html accessed on 07/11/2008