Routing & Router Configuration

A Report to Critically Compare a Number of Routing Protocols; Including RIP v2, EIGRP & OSPF

Paul McDermott

CCNA 2

Table of Contents

1.0 Abstract 3
2.0 Introduction 4
3.0 Protocol overview 5 3.1 RIP v2 Overview 5 3.2 EIGRP Overview 6 3.3 OSPF Overview 6
4.0 Protocol Comparison 10 4.1 Topology Overview 10 4.2 Protocol Types 10 4.3 Administration Distance 10 4.4 Protocol Tables 11 4.5 Algorithm 11 4.6 Metric 12 4.7 Periodic Updates 12 4.8 Hierarchical / Scalable 12 4.9 Load Balance 13 4.10 Comparison Table 14
5.0 Conclusion 15
6.0 References 16

Abstract

The following report is a critical comparison of three routing protocols; RIPv2, EIGRP and OSPF, detailing the protocol features, as well as their similarities and differences. The report takes an in-depth look at the technical elements and algorithms used in these protocols, such as Bellman Ford, DUAL, and the Dijkstra Algorithm; and how these algorithms are used to calculate the routing metric.

The report also discusses the fact that EIGRP is the most desirable protocol to use on Cisco based routers, while OSPF can be used across different router manufacturers.

While looking at the technical considerations that are needed in choosing a routing protocol for a desired network the report will also look into the CPU/memory requirements, and how difficult the protocol is to install and maintain on the network; with RIPv2 being the simplest to run, and OSPF being the most complicated to install and maintain.

Introduction

The following report will investigate the three main routing protocols used and taught in the second module of the CCNA course. This report will investigate the major technical considerations required when deciding on a routing protocol for a network, and how these protocols can be classified. The report will explain how routing protocols function on a network, and also expand on the technical features of the three main protocols discussed – RIPv2, EIGRP and OSPF. The report will use diagrams and tables to compare and contrast the aforementioned routing protocols.

In order for a network to have any form of path redundancy or scalability, a routing protocol must be implemented onto a network. Without this, the maintenance of static routes would almost certainly become taxing on IT resources upon any faults.
Having a routing protocol implemented provides routers the ability to dynamically react to changes in the network topology, and in some cases they can even increase the network performance using load balancing techniques.

Designing a network with the wrong routing protocol could mean that routers in the network would not function at all. For example, EIGRP can only operate on Cisco routers; whereas OSPF & RIPv2 would be the viable choices when non-Cisco routers make up part of the physical network topology.

Routing protocols can be classified into the following

Distance vector – These protocols determine the best past to a remote network by judging point to point distance.
Link state – These protocols use updates to determine the shortest path first in the network.
Hybrid – Hybrid protocols use aspects of both Distance Vector and Link state protocols to determine the best path to a remote network.

Many considerations need to be made when choosing a suitable routing protocol during network design. Factors such as the network size, physical topology, and existing hardware (router model and manufacturer), should all be taken into consideration.

Protocol overview

1 RIP v2 Overview

RIP v2 was one of the first distance vector protocols ever developed, and is still used today in smaller flat-type networks. The current standard for IP version 4 is RIP v2, which includes some significant enhancements over the older v1 allowing it to be useful in today’s modern networks.

RIP is a classless routing protocol that now supports the use of subnet masks in its routing updates; as well as additional features such as authentication, triggered updates, load balancing, and multicast routing exchanges. These additional features over the previous version – and the fact that RIP is very simple to configure compared to other protocols – is the main reason that network administrators are still choosing to deploy this protocol on simple network designs.

RIP uses a metric of hop count and an algorithm known as Bellman Ford to calculate the best path through a network. An example of how RIP routers use hops counts can be seen from the below Fig 1.

[pic]
Fig 1: Taken from Chapter 1.4.5.2 of the CCNA 2 Course.

All packets going from PC1 to PC2 would have a hop count of .2; and the packet would start at router R1 and pass through two routers (R2 and R3) in order to reach the network containing PC2.

RIP uses the routing table to store all its known routes about directly connected networks, and RIP discovered networks. An example of the routing table network shown on Fig 1 would be as below:

R 192.168.3.0 / 24 [120/1] via 192.168.2.2, FastEthernet0/1
R 192.168.4.0 / 24 [120/2] via 192.168.2.2, FastEthernet0/1
C 192.168.1.0 / 24 is directly connected, FastEthernet0/0
C 192.168.2.0 / 24 is directly connected, FastEthernet0/1

Each protocol has a unique code next to its network address. In the case of RIP this code is displayed as the letter R. Another significant value from the routing table is [120/2]; this field shows both the metric (hop count) to a network, and the administrative distance for a network. The administrative distance can be viewed as how trustworthy a routing protocol is. With RIP this is always a value of 120.

When RIP is configured on a network, it will send its routing updates to all adjacent routers every thirty seconds, regardless of whether a change has taken place on the router or not. Although RIP does support triggered updates upon any change, this can have an adverse effect if the trigged updates overlap with the thirty second periodic updates. Thirty seconds may not seem like a significant amount of time, however in networking terms a network is not fully converged until all routers have consistent routing tables. Using the RIP protocol, a network could take up to a few minutes to fully converge as you can see an example of this below in Fig 2.

[pic]
Fig 2: Reconstructed from 4.3.3.1 of the CCNA 2 Course.

During these periodic updates, overlapping triggered updates could create an infinite loop on a RIP network. An infinite loop would be a scenario where a router passes a packet around an incorrectly configured network, resulting in the packet having no final destination, therefore consequentially bouncing between two or more routers. RIP does contain a few additional mechanisms that assist to prevent its routers from being updated with incorrect information; one is the use of hold-down timers, which means a router cannot update its path to a failed network until the timer has expired or a better route has been found. Another way it does this is by using a rule called split horizon, which prevents a router from advertising a network out of the same interface that it received information regarding that network from.

2 EIGRP Overview

Enhanced Interior Gateway Routing Protocol (EIGRP) was first released in 1992 as an enhancement of Cisco Interior Gateway Routing Protocol (IGRP) that was developed in 1985. EIGRP was developed as a classless version of IGRP.

EIGRP is a classless upgrade of IGRP, and like IGRP it is a Cisco proprietary protocol that can only be used on Cisco routers. In the past the term "hybrid routing protocol" has been used to define EIGRP; but Cisco no longer uses this term as it is solely a distance vector routing protocol, even though it possesses the features of a link-state protocol.

Being a classless protocol, EIGRP includes the subnet mask in the route advertisement and supports variable length subnet masking (VLSM). The advantages of this are that fewer IP addresses are needed, and summarization can be controlled within the routing protocol

EIGRP uses less bandwidth than other routing protocols as it does not use periodic updates, rather it uses bounded and partial updates.

EIGRP uses reliable transport protocol (RTP) for the delivery and receipt of EIGRP packets. It can send these packets multicast or unicast, but has the address of 224.0.0.10 reserved for its multicast packets. Although its name suggests otherwise, RTP includes both reliable and unreliable delivery of EIGRP packets, an example of reliable is shown below in Fig 3.

[pic]
Fig 3: Taken from Chapter 9.1.4.2 of the CCNA 2 Course.

The main features of EIGRP are 100% loop free topology, fast convergence, classless routing, and that it uses bandwidth and delay as metrics, and also uses load balancing.

EIGRP uses the DUAL (diffusing update algorithm) as opposed to the bellman-ford or ford-faulkerson algorithms. The use of DUAL means that backup routes of the main route are installed (feasible successor routes), and DUAL also guarantees loop free routes as no hold down times are used.

EIGRP is good for supporting routing in very large networks as it supports the use of multiple autonomous systems (AS) on a single router; this is why EIGRP is the most preferable routing protocol to use on Cisco routers.

3 OSPF Overview

OSPF (open shortest path first) was originally developed in 1987 by the IETF (internet engineering task force) as a link state routing protocol replacement for RIPv1, with OSPFv3 being published in 1999.

OSPF is a very good open protocol that can be used on both Cisco and non-Cisco routers and is comparable to EIGRP due to its fast convergence, redundancy, and scalability in much larger networks. OSPF is a link state protocol which uses five different Link-State Packets (LSP), each serving a specific purpose in the routing process; these are sent using the multicast address 224.0.0.5.

The five types of OSPF Packets this routing protocol use are as follow:

• Hello Packets - Used to maintain and establish adjacency with the other routers running the OSPF protocol. • Database Description (DBD) - Contains an abbreviated list of the sending routers link-state database, and is used by the other receiving routers to check against the local link state database. • Link State Request (LSR) - These packets can be sent by a receiving router to request more information about any entry in the DBD. • Link State Update (LSU) - These are the packets sent as a reply to the above LSR and also to announce new information. • Link State Acknowledgement (LSA) - These packets are sent to confirm the receipt of an LSU.
(Adapted from Chapter 11.1.3.1 of the CCNA 2 Course)

OSPF uses Dijkstra's shortest path first (SPF) algorithm to create an SPF tree. From this tree, OSPF can populate the IP routing table with the best path to each network.

OSPF uses bandwidth as its cost metric, and has an administrative distance (trustworthiness or preference) of 110.

OSPF operates by using router ID's to identify each router on the OSPF routing domain. The router ID is an IP address, which can by statically set on each router by using the IP address configured with the OSPF router-id command or dynamically set. If the router-id is not configured then the router will chose the highest IP address of any of its loopback interfaces, and in the case of no loopback addresses being configured then it will chose the highest active email address of any of its physical interfaces.

The show IP protocols command can be used to verify the router-id, as shown below in fig 4a and 4b.

[pic]
Fig 4a: Taken from Chapter 11.2.4.2 of the CCNA 2 Course.
[pic]
Fig 4b: Taken from Chapter 11.2.4.2 of the CCNA 2 Course.

The router-id is used to select the designated router (DR) and the backup designated router (BDR). These types of routers stop networks becoming flooded by requests for information when the routers are all within the same broadcast domain.

Protocol Comparison

1 Topology Overview

The following topology layout has been created in packet tracer for use in the comparisons below. The gateway router has been configured with all three routing protocols, and has received complete routing updates from all other routers.

[pic]
Fig 5: Created within Cisco Packet tracer, with knowledge from the Cisco course.

2 Protocol Types

OSPF is a link state (LS) protocol; whereas RIP and EIGRP are both distance vector (DV) protocols. Both types of protocols have very different ways of learning about all other routers on their networks; however, they do eventually provide full convergence. DV protocols learn about all remote networks from other routers; this is known as routing by rumour. These are relatively simple to configure, and suitable for flat network designs; whilst LS protocols are more complicated as they create a topology view of the entire network, and know all about other routers and pathways. This LS process allows for a much faster convergence than standard DV protocols.

3 Administration Distance

All three routing protocols have a different administration distance (AD) as defined by Cisco. EIGRP which has an AD of 90 is considered to be more trustworthy than OSPF and RIP, which have an AD of 110 and 120 respectively; however, it could be argued that this viewpoint is biased due to Cisco valuing its proprietary protocol above all others. In the unlikely event that all three routing protocols were configured on the network, only the routes learned by EIGRP would be entered into the routing table. When considering the sample network topology which has the gateway router running all three protocols, it is evident from the routing table below the different AD each protocol has:

RIP = R 192.168.1.0/24 [120/1] via 10.1.0.2, 00:00:22, Serial0/0/0
EIGRP = D 192.168.2.0/24 [90/2172416] via 10.2.0.2, 01:00:54, Serial0/1/0
OSPF =O 192.168.3.0/24 [110/65] via 10.3.0.2, 00:55:56, Serial0/0/1
Fig 6: Taken from Cisco Packet tracer Gateway router.

4 Protocol Tables

Unlike RIP which only has one table, the routing table; EIGRP and OSPF both have contain additional databases or tables they store route information within. EIGRP contains a neighbour table that holds information such as interface types and neighbour address to all directly connected networks (Fig 7a), plus a topology table which contains all routes advertised by its neighbouring routers (Fig 7b). OSPF also contains a neighbour table which provides additional information, such as its neighbour name (ID) and the state of its links (Fig 8a); plus a link state database that contain the cost for each route in the network topology, this can be seen by looking at the OSFP interfaces (Fig 8b) located below.

H Address Interface Hold Uptime SRTT RTO Q Seq
0 10.2.0.2 Se0/1/0 14 01:16:15 40 1000 0 5
Fig7a – EIGRP Neighbour Table from Gateway router.

P 10.1.0.0/30, 1 successors, FD is 2169856 via Connected, Serial0/0/0
P 10.2.0.0/30, 1 successors, FD is 2169856 via Connected, Serial0/1/0
P 10.3.0.0/30, 1 successors, FD is 2169856 via Connected, Serial0/0/1
P 192.168.2.0/24, 1 successors, FD is 2172416 via 10.2.0.2 (2172416/28160), Serial0/1/0
Fig7b – EIGRP Topology Table from Gateway router.

Neighbor ID Pri State Dead Time Address Interface
192.168.3.1 0 FULL/ - 00:00:32 10.3.0.2 Serial0/0/1
Fig8a – OSPF – Neighbour Table from Gateway router.

Process ID 1, Router ID 10.3.0.1, Network Type POINT-TO-POINT, Cost: 64
Fig8b – OSPF – Link State Cost from Gateway router.

5 Algorithm

All three protocols use different algorithms to maintain their routing tables and provide the best path to other networks. RIP uses Bellman Ford, EIGRP uses Diffusing Update Algorithm (DUAL), and OSPF uses Dijkstra Shortest Path First (SPF) algorithm. The Bellman Ford algorithm is the oldest out of the three, and is not up to the same standard as the other two. It is not uncommon for this algorithm to create routing loops on a network, and full convergence can take up to several minutes to complete. DUAL and SPF on the other hand, have a very fast convergence rate; and with the use of backup routes within EIGRP and SPF’s complete topology database, provide a complete loop free network.

6 Metric

Both OSPF and EIGRP use a metric based upon line bandwidth to determine the best pathway to another network, whereas RIP uses a metric based upon hop distance between routers. RIP’s method of calculating the metric is very much outdated, due to the increase in bandwidth links in today’s networks. Even if there is a quicker route to its destination network, RIP would only select the route with the least number of hops to enter into its routing table. The metric for OSPF and EIGRP can dynamically change if there is a bandwidth change to the topology, and in EIGRP’s case this can be much more redefined by setting various other weights within the protocol configuration as seen in Fig 9.

[pic]
Fig 9: Taken from Chapter 9.3.1 of the CCNA 2 Course

The metric for each learned network for all three protocols can be viewed within the routing table, as per Fig 10.

RIP = R 192.168.1.0/24 [120/1] via 10.1.0.2, 00:00:22, Serial0/0/0
EIGRP = D 192.168.2.0/24 [90/2172416] via 10.2.0.2, 01:00:54, Serial0/1/0
OSPF =O 192.168.3.0/24 [110/65] via 10.3.0.2, 00:55:56, Serial0/0/1
Fig 10: Taken from Cisco Packet tracer Gateway router.

7 Periodic Updates

RIP and OSPF are the only protocols which send out periodic updates; EIGRP does not as it is confident enough in its own design to work solely by using triggered updates. Although OSPF does use periodic updates, these are only sent once every thirty minutes to verify all routes are still active; this is referred to as a paranoid update and has no real effect on an OSPF network. RIP sends out its complete routing table once every thirty seconds, and this is crucial to the running of this protocol; however, it is also one of the major downfalls due to the overhead it produces on the network. Both the periodic updates and the network discovery packets for all three protocols are only sent using multicast network addresses; this is one ways the routers keep their updates contained, and do not flood the network with broadcast packets.

8 Hierarchical / Scalable

OSPF is the only protocol that is hierarchal; one of the main features of a LS protocol, and naturally this it the ability to be scalable as well. However, this does not mean that it is anymore scalable than EIGRP, which has a flat design by its DV nature. RIP on the other hand is only suitable for small networks with up to fifteen routers maximum, due to a limitation with its metric algorithm. Both OSPF and EIGRP take completely different approaches in order to be as scalable as today’s networks require.

OSPF can be split into zones known as areas; this helps the network to grow, by keeping the routing process within each area segregated. Without this hierarchical design the OSPF tables would become too large to be efficient on a large network. EIGRP has a very simple approach to being scalable, as its routing updates are only sent to other routers on a need to know basis. The use of manual summarisation on each EIGRP router, in addition to auto summarisation at major boundary routers, allows EIGRP to continue to grow, so long as the network design is efficient.

9 Load Balance

All three protocols are able to provide an element of load balancing across equal cost links from one router to another. RIP can provide this balance on up to six paths, while OSPF and EIGRP can provide equal cost on up to sixteen paths. When considering the below topology for load balance from the source to destination routers, each protocol deals with this in different ways.

[pic]
Fig 11: Created within Cisco Packet tracer, with knowledge from the Cisco course.

RIP would see all three routes from the source to the destination network as the same metric, and load balance all packets equally across all three routes. This would actually decrease the routing performance, as packets travelling across the 64k link would hold up all other routes, and essentially make all three routes perform like three 64k links.

OSPF and EIGRP would only use the path via Route A and Route B for load balance, as their metric is based upon bandwidth, this would give a link speed of 1024kb from the source to the destination networks.

EIGRP does have one feature that no other protocol contains, known as unequal load balancing. This is a method of splitting packets across different metrics, as EIGRP can be certain the unequal paths are loop free. In this scenario, EIGRP could use all three links, gaining a total link speed 1088kb.
10 Comparison Table

| |RIP v2 |EIGRP |OSPF |
|Type |Distance Vector |(Enhanced) Distance Vector |Link State |
|Administration Distance |120 |90 Internal, 170 External, 5 Summary |110 |
|Tables |Routing Table |Routing Table |Routing Table |
| | |Neighbour Table |Neighbour Table |
| | |Topology Table |Link State Database |
|Algorithm |Bellman Ford |DUAL |Dijkstra - Shortest Path First |
|Metric |Hop Count (Max 15) |Cost (Bandwidth + Delay) |Cost (100^8 / Link Bandwidth) |
|Hierarchical Design |No |No |Yes |
|Scalability |15 Routers Max |224 Hops |Very Well |
|Auto Summarization |Enabled by Default |Enabled by Default |N/A |
| | |Can Summarize at each router? |Summarize With Areas |
|Load Balance |Equal Cost up to 6 paths |Unequal Cost up to 16 paths |Equal Cost up to 16 paths |
|Periodic Updates |30 Seconds |None |30 Minutes Link state refresh |
|Network Discovery |Multicast - 224.0.0.9 |Multicast - 224.0.0.10 |Multicast - 224.0.0.5 - All Paths |
| | | |Multicast - 224.0.0.6 - DR/BDR |
|Standards |Open Standard |Cisco Proprietary |Open Standard |
|CPU / Memory Requirements |Low |Medium |High |
|Install & Maintain |Simple |Hard |Complicated | Conclusion

This report has provided an overview of the three main routing protocols - RIPv2, EIGRP, and OSPF; detailing their operation and classification. It has provided a critical comparison of the technical features of the protocols, and given an insight into the drawbacks and benefits that should be considered when deciding on a routing protocol for a network. It has then summarised the major features of these protocols into a comparison table, providing the reader with a quick reference point for each technical feature.

From the protocols that have been compared within the report, it is evident that EIGRP and OSPF are far superior to RIPv2. Having considered the facts, this report concludes that the only reason for installing a RIPv2 protocol on any network would be for the purpose of training with relation to older routing hardware. Due to the lack of scalability with this protocol and its simple metric algorithm, there is no longer any place for this protocol within today’s networks. EIGRP and OSPF are both superior protocols, both of which would provide sufficient routing for any network.

EIGRP would be the protocol of choice on any network running Cisco routers as its proprietary. One of the main features this protocol has over OSPF is Unequal load balancing. Corporate networks can consist of a variety of link types at differing speeds, and so maximising the bandwidth throughout on all links would be desirable.

The report concludes that OSPF would be the protocol of choice for all other networks; with the appeal of this protocol being the fact that it is open standard, and therefore can run on routers developed by other manufactures. This promotes competition between Cisco’s EIGRP, which will in the longer term serve to improve the function of both protocols.

References

Cellsoft. (2004) Internet Routing, Overview & RIP. [Online] Available: http://www.cellsoft.de/telecom/internetrouting.htm [Accessed: 2 February 2014].

Cisco. (2005a) IP Routing, Introduction to EIGRP. [Online] Available: http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080093f07.shtml [Accessed: 2 February 2012].

Cisco. (2005b) IP Routing, OSPF Demand Circuit Feature. [Online] Available: http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094a8f.shtml [Accessed: 18 February 2014].

Cisco. (2014) Course Material for CCNA2: adapted and taught by University of the West of Scotland.

Doyle, J. and Carroll, J. D. (2009) Routing TCP/IP, Volume II (CCIE Professional Development). England: Cisco Press.

Lammle, Todd. (2013) CCNA: Routing and Switching Study Guide: Exams 100-101, 200-101, 200-120. . Sybex Publishing.

Odom, W. and McDonald, R. (2006) Routers and Routing Basics CCNA 2 Companion Guide (Cisco Networking Academy). England: Cisco Press.

SourceDaddy. (2012) Selecting a Routing Protocol - Networking Tutorial. [Online] Available: http://sourcedaddy.com/networking/selecting-routing-protocol.html [Accessed: 25 February 2014].