19
CHAPTER 1 INTRNET ROUTING: AN INTRODUCTION The Border Gateway Protocol (BGP) [Rekhter and Li (1995)] is considered as a backbone of the Internet. It is a protocol which is mainly used in inter-Autonomous System (AS) routing. It delineates the communication between inter-domain routers in the Internet. The main function of a router, which is running the Border Gateway Protocol, is to exchange network reachability information with directly connected peers which are also running the Border Gateway Protocol. 1.1 ROUTER device that is used to forward packets coming from one network destined for another network is called a router. A router is connected with two or more lines each connecting a different network. When a packet is received on one of the lines, the router checks the destination address information in the header of the packet to know the location of the destination network. Then, using the information available in its routing table, it forwards the packet onto the appropriate line which leads to the destination network directly or indirectly. This way the router does its job of directing traffic on the Internet. A data packet is typically forwarded from one router to another through the networks that constitute the internetwork until it reaches its destination node. Routers are used to provide connectivity in networks ranging from connecting nodes in small enterprise networks, large enterprise networks, to large Internet Service Providers (ISPs) networks together. A

CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

  • Upload
    lamlien

  • View
    218

  • Download
    3

Embed Size (px)

Citation preview

Page 1: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

CHAPTER 1

INTRNET ROUTING: AN INTRODUCTION

The Border Gateway Protocol (BGP) [Rekhter and Li (1995)] is considered as a backbone of

the Internet. It is a protocol which is mainly used in inter-Autonomous System (AS)

routing. It delineates the communication between inter-domain routers in the Internet. The

main function of a router, which is running the Border Gateway Protocol, is to exchange

network reachability information with directly connected peers which are also running the

Border Gateway Protocol.

1.1 ROUTER

device that is used to forward packets coming from one network destined for another

network is called a router. A router is connected with two or more lines each

connecting a different network. When a packet is received on one of the lines, the

router checks the destination address information in the header of the packet to know the location

of the destination network. Then, using the information available in its routing table, it forwards

the packet onto the appropriate line which leads to the destination network directly or indirectly.

This way the router does its job of directing traffic on the Internet. A data packet is typically

forwarded from one router to another through the networks that constitute the internetwork until

it reaches its destination node. Routers are used to provide connectivity in networks ranging from

connecting nodes in small enterprise networks, large enterprise networks, to large Internet

Service Providers (ISPs) networks together.

A

Page 2: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

2 | P a g e

A router performs its task in two stages. These stages are often called planes as forwarding plane

and control plane [Khosravi and Anderson (2003) ].

Forwarding plane: A router forwards packets received on an incoming interface to one of

the selected outgoing interfaces. Outgoing interface selection is made based on the

information read from the header of the incoming packet and the information stored in its

routing table.

Control plane: A router makes a routing table to know which route should be used to

forward an incoming packet, and also to select the corresponding physical interface. A

router does this with the help of internal pre-configured addresses, called static routes.

To be specific, a router has the following attributes [Macfarlane (2006)]:

A processing unit.

An operating system.

Network interfaces for receiving/ forwarding packets.

A routing table that indicates which, interface the packets should be

forwarded on to.

Memory to store the operating system, the routing tables, and the

configuration information

A router hardware configuration is similar to a computer’s hardware. The components of a

generic router are shown in figure 1.1, though different vendors’ may have slightly different

hardware.

The common components are the network controllers, the common system bus, central

processing unit (CPU), the bus interface, random access memory (RAM), the flash memory, the

non-volatile random access memory (NVRAM), and the read only memory (ROM). Apart from

Page 3: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

3 | P a g e

the above mentioned hardware, a router has specialized software to make forwarding decisions

and to calculate the best route to every destination prefix.

Figure 1.1: Hardware Details of a Router

The routing software implements various routing protocols in it. These routing protocols

function in distributed manner and coordinate with other routers in the neighborhood.

1.2 ROUTING PROTOCOLS

Routing protocols help routers to communicate with their neighbor routers, exchanging

information that enables them to compute the best routes to any given network prefix. The type

of algorithm used in the protocol determines the type of routes will be preferred. Initially each

router has information of networks which are attached to it directly. Later when they start

sharing this information with their immediate neighbors, then the information reaches to all the

routers in the network. This way every router acquires the knowledge of the location of each

destination prefix, and the information about the topology of the network.

Page 4: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

4 | P a g e

The routers maintain lookup tables to store entries of the routes to reach destination networks,

these routes are classified into two categories first category is of static routes and the second

category is of dynamic routes. Static routes are manually configured by the network

administrator, and dynamic routes not configured by the network administrator rather they are

computed by the routing protocol itself.

Figure 1.2: Interior Gateway Protocol in A University Network

The routing protocols are also divided into two broad categories, first Interior Gateway Protocol

(IGP), second Exterior Gateway Protocol (EGP). Interior gateway routing protocols are used to

forward packets within the networks of an enterprise, a university, or a company. A large

university campus (shown in figure 1.2) has many buildings where each building running a

different network, then to connect these networks together Interior Gateway Protocol is being

used. Exterior Gateway Protocols are used to forward packets between networks of different

enterprises, different universities, or different companies, as shown in figure 1.3. Therefore both

Page 5: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

5 | P a g e

the categories of routing protocols have different areas to work with, hence have different

requirements.

Figure 1.3: EGP Connecting Two Different Organizational Networks

Routing protocols can also be classified based on their delivery schemes. Delivery schemes

could be of four types as mentioned below:

Unicast

Multicast

Broadcast

Anycast

A unicast scheme is the one in which a router delivers the packet to a single destination node

only. In multicast destination nodes are more than one rather it is a group; therefore a packet is

delivered to all those nodes which are members of that group. Broadcast is the method of

delivering a packet to all the nodes irrespective of its address. In anycast, a packet can be

delivered to any of the member belonging to a particular group of potential receivers therefore

whichever member node is topologically nearest on the route that node will be delivered the

packets.

Page 6: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

6 | P a g e

The popular adaptive-routing protocols used as Interior Gateway Protocols are:

Routing Information Protocol (RIP)

Open-Shortest-Path-First protocol (OSPF)

Intermediate System-to-Intermediate System (IS-IS)

For the Exterior Gateway Protocol there is only one protocol, which is:

Border Gateway Protocol (BGP)

Though the Internet has dominance of adaptive routing but the use the non-adaptive routing is

suggested to handle specific issues of networks. However, to configure routing protocols requires

expertise in understanding network complexities and awareness of issues in networks.

Figure 1.4: Network of Routers with Link Cost

1.3 DISTANCE VECTOR ALGORITHMS

Distance vector algorithm is based on the Bellman-Ford algorithm; the people whose

contribution has brought the algorithm are; the Bellman in 1957, Ford and Fulkerson in 1962.

Page 7: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

7 | P a g e

This algorithm was used in the original ARPANET for routing, and at a later stage it was also

used in the Internet for some time under the name Routing Information Protocol ( RIP ). There

are other flavors of distance vector routing like the proprietary protocol from the Cisco the

Interior Gateway Routing Protocol (IGRP ), and the Enhanced Interior Gateway Routing

Protocol (EIGRP ) which are also in use but not as popular as RIP.

In this algorithm the cost, in the form of a number, is assigned to each of the connecting links in

the network, shown in figure 1.4. While selecting the best route to reach any destination a

comparison is done by collecting all the cumulating cost to that destination, then among all the

candidate routes available, whichever offers the lowest cost is selected as the best route to

forward the traffic to that particular destination and similar method is repeated to compute the

best route to each destination in the network. This way each router collects the reachability

information and maintains it in its routing table. These routing tables are updated by exchanging

information, related to the new network destinations or related to the changes in the routes of

previously existing networks, with the neighbor routers.

The distance vector routing requires each router to maintain its routing table indexed, and

contains entries for each router and all the prefixes in the network. This entry has two

components, first the preferred outgoing line to that destination prefix, second, the cost to reach

the network prefix.

The metric used in the distance vector algorithm based protocols might be the number of hops,

delay in milliseconds, queue length along the path etc. The router is supposed to know the

distance to reach each of its neighbors and to reach all the network prefixes of the subnet through

these neighbors. If the routing protocol uses the metric in number of hops, the distance to each

neighbor is just one hop. If the routing protocol uses the metric queue length, the router simply

examines the number of packets in respective queues. If the routing protocol uses the metric

Page 8: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

8 | P a g e

delay, the router can measure the delay directly with special packets called ‘echo packets’, that

the receiver just timestamps and responds back as fast as it can.

1.4 LINK STATE ROUTING

Since the beginning of the ARPANET the distance vector routing was used until 1979, then the

link state routing replaced it. There were two important points responsible for triggering this

change, the first point was the delay metric used in terms of queue length, and distance vector

routing had not considered line bandwidth when computing best routes. This was because of the

fact that in the beginning all the lines were of the same bandwidth that was 56 kbps, therefore

line bandwidth was not considered as an important issue, but after some months when some of

the lines were upgraded from 56 kbps to 230 kbps and others were upgraded to even more as

1.544 Mbps. Then in the changed scenario not considering bandwidth as an important part of the

metric was a major problem.

Though it was also possible that time to modify the existing routing algorithm instead of going

for the completely new routing algorithm. But due to some inherent weakness of long

convergence delay, apart from not taking line bandwidth into consideration. The count-to-

infinity issue, three node instability kind of problems caused the convergence time to be too long

had given enough motivation; therefore it was replaced by an entirely new algorithm, now called

link state routing. There are many variants of the link state routing in use.

The first popular link state protocol is Intermediate System-Intermediate System (IS-IS). The IS-

IS was designed for DECnet, which is a suite of network protocols created by Digital Equipment

Corporation in 1975, and then it was adopted by ISO to use it with its Connectionless Network

Layer Protocol (CLNP).

Since then the IS-IS has gone through many modifications to work with other protocols as well,

most notably the IP. The IS-IS was used in Internet backbones earlier including the NSFNET

backbone. Novell NetWare used a variant of IS-IS called NetWare Link State Protocol (NLSP)

for routing IPX packets. The IS-IS distributes the acquired information of the network topology,

a complete view of topology, to every router. Therefore it becomes easy for routers to compute

Page 9: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

9 | P a g e

the shortest paths to the destination. A router announces its link state information, which contains

reachable network layer addresses. These addresses can be IP, IPX, AppleTalk etc. The strength

of IS-IS is its capability of providing support to multiple network layer protocols at the same

time.

The protocol Open Shortest Path First (OSPF) is also very popular among all the link state

protocols. The OSPF was designed several years after the IS-IS was designed. Then the OSPF

adopted many of the techniques and innovations designed for the IS-IS. Those innovations

include a self-stabilizing method of flooding link state updates, the concept of a designated

router on a LAN, and the method of computing and supporting path splitting and multiple

metrics.

Therefore, as a result, there is very small difference between the IS-IS and the OSPF. The most

important difference is that the OSPF cannot carry multiple protocols’ information

simultaneously, while the IS-IS is designed in such a way that it can carry information about

multiple network layer protocols simultaneously. This feature is very useful in an environment

where multiple network layer protocols are in use.

A router using link state routing performs its task in following steps [Tanenbaum (2002)]:

1. Discover neighbors and learn their network addresses.

2. Measure the delay or cost to reach each of its neighbors.

3. Construct a packet to tell all neighbors about what it has just learned.

4. Send this packet to all other routers.

5. Compute the shortest path to every destination in the network.

The above mentioned steps are shown in figure 1.5, the complete topology and all types of

delays are experimentally measured and distributed to every router. Then the Dijkstra's algorithm

is applied to compute the shortest path to every destination prefix in the network.

The process of learning about neighbors starts when a router is booted; its first task is to know

who the neighbors are, then to exchange the information of the connected networks with all the

neighbors. The knowledge about neighbors is acquired by sending a special packet called a

HELLO packet to each directly connected neighbor. The neighbor routers are also supposed to

Page 10: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

10 | P a g e

send back similar replies containing their known information. These routers` identification must

be globally unique because when a distant router later hears about these routers it must be able to

identify correctly.

Figure 1.5: Functioning of Link State Protocol

After the Interior Gateway Protocols, it is time to focus on Exterior Gateway Protocol. There are

very few protocols available to function as an Exterior Gateway Protocol, today the choice for

Exterior Gateway Protocol is the Border Gateway Protocol (BGP) only.

1.5 BORDER GATEWAY PROTOCOL

The Interior Gateway Protocols mentioned in earlier sections are capable of taking care of the

connectivity inside the networks of an organization, university, enterprise etc., each of these

networks belonging to a separate body ( organization, university, enterprise ) are managed by a

Page 11: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

11 | P a g e

single administrative team, known as Autonomous Systems (AS). The interconnection of

multiple ASes forms the Internet.

The Internet is composed of millions of networks and of thousands Autonomous Systems (shown

in figure 1.6). The Internet was not always so large in size. In 1980s the Internet was still the

ARPANET, till that time there was no separation of networks in Autonomous Systems (ASes).

Those days the ARPANET was one big internetwork under a single routing administration. A

protocol which was a distance vector routing protocol called the Gateway-to-Gateway Protocol

(GGP) used to route traffic in the ARPANET. Every router in the ARPANET had to maintain a

route to every destination network. Therefore GGP was basically an interior routing protocol.

Initially, GGP was the protocol of choice, but Routing Information Protocol (RIP) soon took

over as the default interior routing protocol of the time. Then the IS-IS came into use, which was

then followed by OSPF [Macfarlane (2006)].

Figure 1.6: Internet Map [courtesy ‘the opte project’, [Opte (2013)]

Therefore when the size of the networks in ARPANET grew up, it was realized that there was a

need to divide each big network of an organization in one Autonomous System. Because the

Interior Gateway Protocols had a severe limitation of scalability. Therefore the only choice left

Page 12: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

12 | P a g e

was to confine an internetwork into one unit of the larger network. When there was an issue of

connecting two or more internetworks, which are from different organizations ( different ASes )

then Exterior Gateway Protocols were used for the purpose.

The first protocol to provide routing between Autonomous Systems was the Exterior Gateway

Protocol (EGP). The EGP made a provision to assign an Autonomous System Number or AS

Number (ASN) to each group of internetworks belonging to one organization, this AS number

(ASN) was used to uniquely identify it. Therefore the routing between the Autonomous Systems

was handled by EGP. Within a large network which is now known as an AS, the local

administrators were free to use the any routing method or protocol for their internal networks to

route traffic. The EGP was used initially but soon the Border Gateway Protocol took over the

role of EGP.

Although the deployment of the Exterior Gateway Protocol (EGP ) for connecting Autonomous

Systems (ASes) solved several problems, but soon it was realized that it had many limitations,

like it required an administrator to manually configure all the routes. The EGP also lack of a

native mechanism to determine the optimum of routes or a mechanism to check and prevent

routing loops. The reason for not having these features was to have no consideration for

connecting with multiple backbones. Then in 1989, Border Gateway Protocol was first released

by Internet Engineering Task Force popularly known by its acronym IETF [ Lougheed and

Rekhter (1989)]

Since 1989, the BGP has proven its worthiness as a solution for Interdomain routing. It has gone

through many revisions, the current version is 4 [Rekhtar and Li (1995)]. Internal BGP (iBGP)

can be used in large networks of an organization to forward traffic between subdomains.

External BGP (eBGP) is used to forward traffic to and from the organization’s networks and to

participate in global Internet routing.

Unlike Interior Gateway Protocols the BGP works in a different way. For example, the BGP

does not advertise only network links rather it advertises the complete path to the destination

prefix. The BGP always keeps the complete path to the destination prefix by keeping a list of

Autonomous Systems (ASes). Therefore it is called as a path vector protocol. The BGP uses a

Page 13: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

13 | P a g e

variety of attributes to each path it learns over the AS networks. It uses these attributes to

implement various policies in traffic forwarding. The routing policies facilitate the administrator

the control to decide what prefixes are to be distributed between ASes.

1.6 BGP WORKING

The BGP uses the Transmission Control Protocol (TCP) for the connectivity, because the TCP is

a reliable protocol. The BGP uses five different types of messages for exchanging information

with neighbors, as shown in table 1.1.

Table 1.1: Type of Messages in BGP

TYPE MESSAGE PURPOSE

1

Open

It is used to start a BGP session with other BGP routers. With

this message a router introduces itself, exchanges optional

BGP features, and its capabilities.

2

Update

This message is used to advertise routing information to other

BGP routers.

3

Notification

It is used to report an error. It is primarily used to reject an

open message, or to report a problem with an update message.

4

KeepAlive

This message is exchanged on the BGP session when there is

no other traffic to allow the BGP routers to distinguish

between a failed connection and a BGP peer that has nothing

to say ( that is , no new routes to advertise).

5

Route-Refresh

This message type is used as a specific request for a BGP

router to re-advertise all of its routes in its routing table using

Update messages. This message was not defined in the

original BGP-4 RFC (RFC 1771 ), it but was added in RFC

2918.

Each of the message type described in above table 1.1 starts with a standard header consisting of

three fields; the marker, the message length, and the type of a message field. The Marker field is

to carry authentication data. Open message must contain data in hexadecimal format. The length

Page 14: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

14 | P a g e

field contains the length of the entire BGP message, and the message type field of the header

contains information about the type of message being carried by the packet. The format of the

BGP message header is shown in figure 1.7.

Figure 1.7: BGP Message Header

The BGP router which speaks to another router of different AS is called BGP speaker. All

aforementioned messages build different types of databases in the form of tables. The BGP

speaker router acts as the gateway and distributes information to all the routers in the AS.

1.7 MOTIVATION FOR THE WORK

The BGP is being used now days by every Internet Service Provider (ISP), Organizations,

Companies, etc. A series of additional requirements have emerged and the BGP has been

extended to address them. The usages of the BGP have exposed many limitations and the

operational experience help realizing the need to address the problems properly and timely, so

that the Internet remains available without compromises. The area of Internet routing is of prime

importance because of the fact that the BGP has no substitute till date. The routing protocol ( the

BGP) used on the Internet does not offer optimality of routes. Its only purpose is to provide

reachability and stability in the global routing system. In ISPs and corporate enterprises the

needs to run BGP on their networks are profit oriented. An intermediate service provider with

whom an enterprise has no business relationship do not care if the enterprise’s traffic follows

Page 15: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

15 | P a g e

suboptimal routes, nor do they have any incentive to do so. Therefore if the protocol is not

modified or changed to accommodate the newly arisen issues, the entire system which has a

dependency on the Internet will become vulnerable. As of today we are not in the position to

afford the disruption of the Internet and its services.

1.8 ORGANIZATION OF THE THESIS

From the study of previous work in literature survey, it is observed that the many developers

have contributed to the successful and efficient functioning of the Internet. Because of the

continuity in contribution from the IETF and others, the BGP has experienced many revisions so

far. But some major issues are still present (as mentioned earlier), which must be addressed to

keep the Internet running efficiently in order to get full advantage of the technology. Lots of

work has been reported by authors in the literature to overcome those limitations. This thesis

presents the analysis of all these major issues of Internet routing with their comparative study,

followed by proposed solution to each and every problem.

The work reported in the thesis is organized in six chapters, as given below –

Chapter-2: Literature Survey

This chapter presents an in-depth literature survey of the Internet routing using BGP and its

issues. The chapter starts with basic routing problems and the manner in which these problems

have been addressed earlier, and then the further issues arisen because of the limitations of

addressing technique.

Many authors have contributed in highlighting the issues in Interdomain routing [Wang et al.

(2009)]. These authors have given their approaches to handle each issue in the routing. These

issues range from initial detection of the existence of the loop in the forwarding path to the

provision of having more than one path in their routing information bases [Elmokashfi et al.

(2008), Abuzneid and Stark (2010), Li et al. (2011), Bates et al. (2006)].

Page 16: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

16 | P a g e

Labovitz et al. have proposed the use of an approach called the Sender-Side Loop Detection

(SSLD). It is a method to decrease convergence time when the link or node failure disrupts the

reachability to the destination [Labovitz et al. (2001)].

The Classless Inter-Domain Routing, which is popularly known as CIDR, was described by

Rekhtar Y, and Li T, in the year 1995 as tool to accommodate the addressing problems [Rekhtar

and Li (1995)].

The Route flap damping was designed to control flaps by Z. Mao et al in the year of 2002.

Whereby flapping routes are suppressed for the relatively long duration until these routes

become stable. Once stability is accomplished which is based on some time interval threshold

suppressed routes are again used [Mao (2002)].

Zhenhai Duan et al. in the year of 2007 put their efforts to identify the flaps and classify the

messages exchanged during a link failure or path withdrawn [ Duan (2007) ].

The detailed study has been carried out in the thesis under the heading of Literature Review in

the second chapter of the thesis.

Chapter-3: Stabilizing the Interdomain Routing Protocol after Failure

In the Internet environment, the Diversity-path-aware Border Gateway Protocol (D-BGP)

increases path diversity by advertising multiple paths. In case of failure, D-BGP, selects the

shortest available alternative path, but does not consider the quality of the path to be selected

after failure. To overcome these limitations, changes are required in the procedure of alternative

path selection which increase the chances of the survival after failure.

In this chapter an approach to stabilize BGP protocol based on the quality of the path has been

proposed. Using the unconventional approach for alternate path selection while the main path

becomes unavailable. Attributes of scalable link state routing have been used for calculating the

link availability and bandwidth availability.

Page 17: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

17 | P a g e

The proposed approach performs better in terms of higher available bandwidth, minimum delay,

and also with minimum packet drop. Therefore it provides stability in the routing during failures

[Kumar and Kumar (2013a)].

Chapter-4: Minimization of table size through Aggregation

BGP maintains reachability information in routing information base (RIB). As the size of the

Internet has grown at a faster rate than expected and the routing table size reaching at alarming

state [BGPReport (2012)].

In this chapter, the issue of aggregation in growing size of the Internet has been addressed, which

is raising concerns among Internet designers as the routing table growth leading to the state of

greater difficulty in managing the huge amount of forwarding information on the relatively

limited memory size of line cards. Aggregation is the method of limiting the volume of data in

the forwarding information base. Most of the existing approaches have shown an aggressive

aggregation tendency [Qing Li et al (2011)] that leaves many problems like black holes,

overburden next hops links, inappropriate path selection etc. in the forwarding information which

subsequently degrades the overall performance of the Internet. In this chapter an approach

AGRIB-BHF to overcome aforementioned problems of aggregation has been developed [Kumar

and Kumar (2012)].

Chapter-5: Network Availability in Interdomain Routing

The Internet experience changes in connectivity due to rapid changes in topologies. The

symptoms of these route instability cause non-availability of those routes which previously

existed in the routing table. Whenever a route is withdrawn from the router’s routing table it

selects the new route if available and advertises the newly selected route to its peers otherwise it

sends a withdrawal of that route to its peers. But sometimes this simple looking process lasts for

the relatively long duration and it ends with the exchange of many updates. Therefore the

network may remain unstable during this period and the convergence is not achieved for a long

duration.

Page 18: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

18 | P a g e

Through this chapter an algorithm to keep the value of the Minimum Route Advertisement

Interval (MRAI) timer variable has been proposed, which is unlike the conventional approach of

keeping it constant. The proposed approach makes the MRAI timer value varying depending on

the network conditions for the path to the prefix advertised. The approach helps the Internet to

reduce the convergence time and makes the network available relatively quicker [Kumar and

Kumar (2013b)].

Chapter-6: Routing in Large Networks Inside Autonomous System

The Internet is the interconnection of different large networks together. The administrators use

Interior Gateway Protocol for routing inside these large networks. Managing routing issues in the

network is a challenging task for administrators although they have a choice of selecting from a

primitive routing protocol like Routing Information Protocol (RIP version 1) to the advanced

routing protocol like Open Shortest Path First (OSPF ) or Intermediate System-to-Intermediate

System (IS-IS) protocol. Irrespective of the above mentioned protocols issue of scalability has

been problematic for administrators. They let their network grow up to the limit where it

becomes unmanageable [ Halabi (1997) ].

This chapter presents the analysis of issues in managing large networks inside Autonomous

Systems. With the help of segmentation through Route Reflectors and Confederations [Cisco

(2012), Bates et al. (2006)] a large network become manageable and improves overall

performance of routing and forwarding. The analysis of the two approaches has been completed

and it helps to overcome the limitations of the large networks [Kumar and Kumar (2013c)].

Chapter-7: Conclusions and Future Work

The last section of the thesis concludes the work and addresses the scope of future

enhancements. The following objectives have been successfully accomplished:

i. Changes made to the default procedure of path selection during failures proved

beneficial. The results show that the number of packet drop is low, and available

bandwidth is more [Kumar and Kumar (2013a)].

Page 19: CHAPTER 1 - Information and Library Network Centreshodhganga.inflibnet.ac.in/bitstream/10603/10640/11/11_chapter 1.pdf · network reachability information with directly connected

19 | P a g e

ii. The method developed for reducing routing table size successfully reduces the table

size. But the reduced size is neither aggressive nor greedy in function, therefore it

successfully remove the black hole before aggregation. The results also show that it

helps utilizing next hops in a relatively balanced manner [Kumar and Kumar (2012)].

iii. Frequency of successive updates exchanged with the peer routers has been controlled

by making minimum route advertisement interval timer dependent on network

conditions. It helps the network reaching the state of convergence quickly [Kumar

and Kumar (2013b)].

iv. Two approaches the Route Reflection and the Confederation to partition the large

network into smaller networks has been analyzed. The results show the advantages of

each approach and precautions to be exercised to overcome the limitations emerged

due to segmentation.

In future the issues of security and authenticity concerned with the routes advertised may be

taken care of to develop the BGP as more secured to survive the attacks on routes.