A Comparison of Centralized Routing Protocol and Distributed Routing Protocol

Song Wei，Li Mengli

（Beijing University of Post and Telecommunication，Beijing，China.E-mail：ee08b139@mail.bupt.edu.cn）

Abstract：In dynamic routing system，a protocol on one router communicates with the same protocol running on neighbor routers.The routers then update each other about all the networks they know about and place this information into the routing table.If a change occurs in the network，the dynamic routing protocols automatically inform all routers about the event.If static routing is used，the administrator is responsible for updating all changes by hand into all routers.Typically，in a large network，a combination of both dynamic and static routing is used.

1 Introduction

Computer networks have become a critical complex structure.In today’s world Network systems achieve interloper ability among heterogeneous systems through the use of a set of well established protocols.The generality and wide acceptance of the TCP/IP protocol suite has significantly contributed to theVast success of the Internet.

In the dynamic route calculate algorithms，there are three main routing protocol，one is centralized routing protocol，another is distributed routing protocol，and also hybrid routing protocol.

In this survey，we will have a general sense of how centralized and distributed routing protocol work and the advantage and disadvantage between them，also some points about how they influence the network.

2 The Centralized Routing Protocol

This centralized routing protocol，in the sense that it requires a completeView of the network in order to optimally allocates a connection request.

The principle of the centralized routing protocol is that there exists a special hardware module，like a server – often router − in the switch board，which is responsible for package retransmission.

Every node of the networks transfer the information of their own to the special hardware module—the network control center，then the module will calculate the request based on the information from every node，and at last give the best routine for the packets which want to be send.

3 The Distributed Routing Protocols

The main idea of the distributed routing is that every node has a routing table，it can calculate the shortest path and it is unnecessary to send the routing information to the monitor and no requirements of a completeView of the network.

There are several algorithms of the distributed routing protocol，in this paper，I choose to important algorithms.

3.1 DistanceVector Routing Algorithm

The main idea of DistanceVector Routing algorithm is that：

Every node communicate with all adjacent node each other periodically by sending information，then the information of Routing refresh by a group（V，D），V means which node（number）can be choose by the route，and D means the distance between the current node and the destination node（how many jump between them）.

The receiver routing refresh the information node recalculate and modify it's routing table.

DistanceVector routing algorithm has the advantages of simple and reliable.However it does not apply the large network which may has a drastic change environment.Since a certain node routing won’t change until fluctuations from adjacent node spreads out to the certain node routing，and the process may be slow，so the changing of the certain node routing suffer from a big delay，it’s called “slow convergence”.Therefore，in the routing refresh process of distanceVector routing algorithm，there may appear some routing disagreement

3.2 Link-state Routing

The basic idea of the link-state routing algorithm is not complex，it can be divided into five sections describe the following：

（1）Each node must find all its neighbors.

When a node starts to work，it will send a special HELLO messages through a point to point link，and through the other end of the link node sends a response packet to tell who you are.

（2）Every node measured the time delay of each neighbors，and other parameters.

Link- state routing algorithm requires each node knows the delay of each neighbors.The most direct way of measuring is send a special ECHO message through the link between the certain node and its neighbor，and require the response message back immediately when its neighbor receive the message.We divide the time we measured by 2 and we got a reasonable estimation.In order to get a more accuracyValue we can do the measurement several times and get the averageValue.

（3）Establish the link-Status messages

After gathering all the information for communication，the next step is to establish the format message which contains all the data for every node.The format message begins with the sender's identifier followed by a sequence number as well as a list of all its neighbor.The information of delay is given for every certain neighbor.

To establish a link-Status message is not complexity，the difficulty is to decide when to create them.One possible approach is to establish them at regular intervals periodically.Another possibility is when the node detects the occurrence of certain important events，establishing them.For example，establish them when the collapse of a link or a neighbor or restored occurred.

（4）Distributed of link-Status messages

The basic algorithm is to use the sequence number distributed of the flooding method.

Since this algorithm is using the sequence number circularly，some node may suffer the crash before，or may be some sequence number was use incorrectly before，it may cause the nodes using different topology structure.It may cause the unstable.In order to avoid the unstable，every format message will contain a TTL，when the time is up，the certain format message is scrapped.

（5）Calculate the new route

Once a node has collected to all the links from other nodes-status messages，it could then construct a complete network topology map，by using the Dijkstra algorithm we can achieve the shortest possible path to all destinations on the local structure.

Link-state routing algorithm is to calculate the shortest path by each node independently.It can adapt to network changes quickly.However，compared to distanceVector routing algorithm，it is more complex and difficult to achieve.

4 Advantages and Disadvantages of Two Kind of Routing Protocol and How They Influence the Network

4.1 Centralized Routing Protocol

From the survey above，we can easily figure out the advantage and disadvantage of the centralized routing protocol.

• The centralized routing protocol is easy to control.
• The centralized routing protocol is easy to achieve and cost less money.
• The capabilities of the network route are limited to the network control center’s capabilities.If the network control center’s capability is high，the network route’s property is high andVice-versa.
• If the breakdown of the network control center occurs then it cause the collapse of the whole network.
• This routing protocol can only suitable for small and low-speed changing networks.If a big and fast changing networks using the centralized routing，this protocol can hardly competent.

4.2 Distributed Routing Protocol

From the survey above，we can easily figure out the advantage and disadvantage of the distributed routing protocol.

• The distributed routing protocol is unnecessary for manually update the routing table，this characteristic can ensure that the distributed routing protocol is especially good for the big and fast changing networks than the centralized routing protocol.
• The distributed routing protocol is time saving and cost saving.
• One of the distributed protocols OSPF based on throughput rate，traffic condition，round-trip time and other parameters to choose the shortest and best routing path in order to reduce the load of network efficiently.
• The distributed routing protocol can achieve different QoS routing service.
• The distributed routing protocol can give a high speed response when there is a linking condition changes.
• The database over flow may happen in the distributed routing protocol.
• The protocols’ normal behaviors may be influenced by changing the IP address.

5 Conclusion

In the survey，we get familiar with the centralized and distributed routing protocol，and also their advantage and disadvantage in different occasion.The different protocols have different performance in different environment.

As a conclusion，there won’t be anything that will suitable for all cases，the only way to choose the routing protocol is choose the protocol that fit for the requirement environment.

Reference

[1] http：//baike.baidu.com/view/4303326.htm

[2] Internetworking With TCP/IPVolume I Principle，Protocol，and Architecture（Fifth Edition），Douglas E.Comer.

[3] http：//zh.wikipedia.org/wiki/OSPF