Analysis Of Successive Link Failures Effect On Rip

Published Date: 02 Nov 2017

Routing process is the backbone of any Inter/Intra domain scenario. Several routing protocols exist nowadays but out of them the most commonly used are Routing Information Protocol (RIP) and Open Shortest Path First (OSPF). The prime objectives of this project are to investigate the consequences of deploying RIP and OSPF simultaneously on a network and compare the convergence delay of RIP and OSPF using two different media types i.e. Ethernet and Frame Relay. This independent study will introduce the characteristics of IP addressing. The similarities and differences between Variable Length Subnet Mask (VLSM) and Classless Inter-Domain Routing (CIDR) will be reviewed. Moreover, the advantages of the classless, over the class full nature of a routing protocol will be rationalized as well. The process of building of a routing table will also be discussed. The section will end with a detailed examination of convergence time of RIP and OSPF over Point to Point DS3/DS1 links. Experiments involving five Cisco 7200 series routers with heavy load of video traffic flow will be performed. The two cases of interests are: Impact of a failure link to the RIP convergence in three different circumstances and Impact of a failure link to the OSPF convergence in three different circumstances. We will be testing each case with three different Link failure and recovery time periods.

Keywords: RIP, convergence delay, OSPF, Link Failure, Link Recovery, dead timer and VLSM.

Acronyms

Abbreviation and their meanings

AD Administrative Distance

The "trustworthy" of the Inter/Intra domain routing protocol

RIP Routing Information Protocol

Open standard distance vector routing protocol.

OSPF Open Shortest Path First Routing Protocol

Open standard link state routing protocol.

IETF Internet Engineering Task Force

IP Internet Protocol

A routed protocol that works at OSI layer 3

LAN Local Area Network

Campus network

LS Link State

This is a type of intra-domain routing protocol that floods the link state update throughout the area.

PPP Point to Point Protocol

Open standard Point to Point encapsulation protocol

VLSM Variable Length Subnet Mask

The ability to support variable length of subnet masks.

Table of Contents

Title Page Number

Abstract 4

Acronyms 5

Introduction 6

IP Addressing 8

RIP and OSPF 11

Experiments 13

Conclusion 22

References 23

Introduction

The goal of this independent study is to investigate the behavior of routing/network convergence of RIP and OSPF. It begins with an explanation of IP addressing. The report includes topics such as Frame Relay, Variable Length Subnet Mask (VLSM), Classless Inter-Domain Routing (CIDR) and classful versus classless. Next, the report discusses the two routing protocols: Routing Information Protocol (RIP) and Open Shortest Path First (OSPF) into great detail. The report then examines the structure of a routing table and the route selection process.

In order to be practical in the investigation of the routing convergence, we perform an experiment that involved four Cisco 7200 series routers. It is assumed that an end customer requires redundancy for its connectivity across the network. For that we have evaluated following four scenarios of link failures to check network convergence at different time stamps.

Scenario 1: states that we have redundant Point to Point WAN connectivity over DS3/DS1 lines using RIP.

Scenario 2: states that we have redundant Point to Point WAN connectivity over DS3/DS1 using OSPF.

We conduct the experiment such that network convergences under different failure situation are examined. We will evaluate the above mentioned scenarios in the below mentioned three situations/time stamps of link failures and its recovery.

1.1 Topology Layout

Figure 1

IP Addressing

In order to understand any routing protocol, one must have in depth understanding of IP addressing. Hence, we take in a brief discussion of the IP addressing scheme. Next, we discuss the concepts of Variable Length Subnet Mask (VLSM). This is a technique for making IP addressing more efficient. At the end, we cover the bases of classless and classful behavior of a routing protocol.

2.1 IP Address and Subnet Mask

The IP addressing space in North America is administered by the America Registry for Internet Number (ARIN). An IP address is 32 bits in length with two parts: network number and host number. The length of the network number is different for different classes [11].

2.1.1 IP Address Classes

IP address is defined in five classes. They differ in the number of hosts that can be attached to the network/sub-network. The network portion is assigned by ARIN while the host portion is chosen by the network administrator.

When referring to the network portion, the typical nomenclature is to put a "0" in the host octet, e.g., 172.22.0.0. When referring to the host portion, then it is to use the complete address as the host address, e.g., 172.22.50.1.

Class A Network portion. Host portion. Host portion. Host portion

Class B Network portion. Network portion. Host portion. Host portion

Class C Network portion. Network portion. Network portion. Host portion

Class D Reserved for Multicast

Class E Reserved for experimental purposes

Private address range allows the network administrators to create their own network address schemes; these addresses are not routable over public network directly. Their ranges are

Class A 10.0.0.0 – 10.255.255.255

Class B 172.16.0.0 – 172.32.255.255

Class C 192.168.0.0 – 192.168.255.255

2.1.2 Subnet

Subnet is a concept of subdividing that extends the network number one step further. Let’s consider a network administrator is given any of the class of IP address block. It is required to divide the host portion into different sub-networks in order to separate and manage the traffic streams. By using the concept of sub-netting, the network administrator can decide on the size of the subnet block as per requirements.

A subnet mask is used to identify the host and network boundary. It uses binary ‘1’ to denote the network and subnet bits, and binary ‘0’ to denote the host bits. For the host address 172.22.1.4 with a subnet of 172.22.1.0, the subnet mask is:

11111111.11111111.11111111.00000000 (or 255.255.255.0 in decimal)

Binary representation

Example of an IP address and the subnet mask:

IP address: 10101100 00010110 0000001 00000100 = 172.22.1.4

Subnet address: 10101100 00010110 0000001 00000000 = 172.22.1.0

Subnet mask: 11111111 11111111 11111111 00000000 = 255.255.255.0

A short form is used to show IP scheme in a more summarizing style. The number of ones in the subnet mask is mentioned after the IP address, proceeding with a slash (/). The above example has 24 ‘1’ in the subnet mask field. Subsequently, the IP address and its subnet mask can be written as 172.22.1.4/24. This can be stated as a "bit mask".

2.2 VLSM

As the networks grow rapidly over the world and it leads to a shortage of IP address. The main cause is its pre-set subnet mask. Variable Length Subnet Mask (VLSM) is designed to resolve this problem. VLSM divides the standard class into smaller subnets

2.2.1 Variable Length Subnet Mask (VLSM)

Suppose that it is required to divide a class C network into three subnets: one consists of 90 hosts and two consists of 40 hosts each. Even though a "class C" network has 254 host addresses available, this cannot be done by simply divide the "class C" address space into two 127-host network bits or four 63 host network bits. The only solution is to split the entire address space into two big blocks, each with 127 host addresses, and further sub divide one of the blocks into further two smaller blocks, each with 63 host addresses.

This method of dividing the IP address block into different sizes is called Variable Length Subnet Mask (VLSM). It is flexible and can suite the different requirement. It also reduces the waste of IP address [8].

RIP and OSPF

Routing Information Protocol (RIP) is one of the extensively deployed routing protocols. It uses a distance vector algorithm. It is simple to configure, but has a number of disadvantages.

3.1.1 Algorithm

Routers pass periodic copies of their routing table to neighboring routers and accumulate cost. RIP uses hop count as the metric for each link. For example, consider three adjacent routers, A, B and C connected in a straight line. Router A passes its routing table to Router B; Router B adds one to the metric and passes the routing table to its other neighbor, Router C. The same step-by-step process occurs in all directions between direct-neighbor routers [7].

3.1.2 Topology Change

The routing table must be updated whenever there is a change in inter-network topology. A table update process requires each router to send its complete routing table to each of the adjacent neighbor routers. When a router receives an update, it compares the update with its current routing table. It adds the metric of reaching the neighbor router to the path metric advertised by the neighbor router to establish a new metric to reach the advertised destination.

3.1.3 Problems

There are a number of issues relevant to RIP. First, the slow convergence may cause inconsistent routing entries, occasionally results in routing loops. When there is a link failure, other routers cannot receive the failure notification update before advertising their own routing tables. Therefore, the network builds up the incorrect routing table and increments in the metric. As a result the metric can ultimately approach to infinity.

3.1.4 Disadvantages

There are several disadvantages to RIP. The network is restricted to the size of 15 hops due to the solution to the "count to infinity" problem. In addition, the periodic broadcast of the routing table consumes bandwidth. The convergence is generally slow too.

3.2 Open Shortest Path First (OSPF)

Open Shortest Path First (OSPF) was developed by the Internet Engineering Task Force (IETF) as a replacement of the problematical RIP in RFC 2328 [9]. This is a nonproprietary routing protocol for the TCP/IP protocol family with many advantages over RIP [2].

3.2.1 Algorithm

OSPF advertises link-state update packets that contain local information for each router’s connectivity. Each router exchanges local and external link state information/updates and hence able to generate a shortest path tree towards destination. Each router uses this pre-defined topology to calculate the shortest path to each destination. Recalculation occurs only if there is any change in internetwork topology like link failures or node failures.

3.2.2 Topology Changes

Each router keeps track of the link states of its adjacent router. Whenever there is a change, router notifies other neighbors by sending a link-state update packet. Other routers then reconstruct a complete map of the inter-network topology to reach complete state of convergence.

3.2.3 Problems

Unsynchronized updates and inconsistent path decisions are the main problems of OSPF. Routers cannot determine the most recent update when two different link-state updates arrive at approximately the same time. If the link-state packet is not correctly distributed to all routers, invalid routing entries will be resulted. This problem is relatively minor when comparing to the problem encountered by RIP. This can be solved easily by coordinating the updates.[4].

3.2.4 Advantages and Disadvantages

OSPF has both advantages and disadvantages. Some advantages of OSPF are:

It is one of the highest-performance and open standard routing protocols.

It is a classless routing protocol in nature.

It provides better shortest path routing and is fast to fault-discovery and rerouting.

It consumes minimal link overhead when the network is in steady state (less link failures).

It has been validated by the IETF and implemented by many vendors in the market.

Some disadvantages of OSPF are:

It stresses a higher processing and memory requirement as compared to RIP.

It has a large bandwidth requirement at the initial link-state packet flooding.

Experiments

Experiment consists of 4 Cisco 7200 series router. We will consider two different scenarios with both routing protocols as per mentioned below three different link failures time stamps situations.

First Situation

S.No.

Status

Time(Seconds)

1

Fail

240

2

Recover

420

3

Fail

520

4

Recover

580

5

Fail

610

6

Recover

620

7

Fail

625

8

Recover

626

9

Fail

726

10

Recover

826

Table 1

Second Situation

S.No.

Status

Time(Seconds)

1

Fail

240

2

Recover

480

3

Fail

720

4

Recover

960

5

Fail

1200

6

Recover

1440

Table 2

Third Situation

S.No.

Status

Time(Seconds)

1

Fail

30

2

Recover

60

3

Fail

90

4

Recover

120

5

Fail

150

6

Recover

180

7

Fail

210

8

Recover

240

9

Fail

270

10

Recover

300

11

Fail

330

12

Recover

360

Table 3

4.1 Simulation (First Situation)

It states that we have redundant connectivity over DS3/DS1 lines using RIP and OSPF. We are going to configure OPNET to simulate this scenario with first situation (Table 1).

Referring to our network topology in Figure 1 initially we are going to add application utility in the scenario and then configure it with the following parameters.

Figure 2

Then we are going to add Profile utility in which we will refer what we have already configured in the Application utility.

Then we will add Failure and Recovery utility to our network topology and configure it according to First Situation time stamp Table 1.

Finally we are going to configure/execute discrete event simulation for approximately 15 minutes as per mentioned time period for link failure and recovery for both routing protocols.

After completing this phase we end up with the individual and overlaid graphical results for simulation of First Situation.

Individual Average Graph of RIP and OSPF Convergence

Overlaid Average Graph of RIP and OSPF Simulation

4.2 Simulation (Second Situation)

It states that we have redundant connectivity over DS3/DS1 lines using RIP and OSPF. We are going to configure OPNET to simulate this scenario with second situation (Table 2).

All the configuring parameters will be same as of First situation simulation except Failure and Recovery utility and discrete event simulation execution time period, execution time period in this case will be 25 minutes approx.(1440 seconds) Table 2. Results are shown in individual and overlaid graphical form for simulation of Second Situation.

Individual Average Graph of RIP and OSPF Simulation

Overlaid Average Graph of RIP and OSPF Simulation

4.3 Simulation (Third Situation)

It states that we have redundant connectivity over DS3/DS1 lines using RIP and OSPF. We are going to configure OPNET to simulate this scenario with third situation (Table 3).

All the configuring parameters will be same as of second situation simulation except Failure and Recovery utility and discrete event simulation execution time period, execution time period in this case will be 06 minutes approx. (360 seconds) Table 3. Results are shown in individual and overlaid graphical form for simulation of Third Situation.

Individual Average Graph of RIP and OSPF Simulation

Overlaid Average Graph of RIP and OSPF Simulation

Conclusion

It's difficult to find the best practical performance protocol, but regarding to the size of the network you can choose the suitable for your network. By using OPNET simulator I did an experiment to test the convergence time between RIP and OSPF in problematical times and I concluded that RIP is faster than OSPF in convergence, in case the number of routers less than 15. I like to share my experiment with experts to shed some light on it as we generally study that OSPF converge faster that RIP, but in relatively smaller networks i.e. less than 15 routers it is a different case. This could of two reasons, 1) In case of complex topologies; OSPF may send updates faster than RIP which affects other routes. 2) In case of indirect link failures where we need to consider dead time interval, OSPF has low dead time (40) considered to RIP (180). Since RIP has a dead interval of 180 seconds so after link failure this protocol consider itself converged for 180 seconds then it starts rerouting processes but on the other hand OSPF has low dead time interval so OSPF will go through convergence three times more as compared to RIP hence resulting in slow convergence.