Application of Mobile Agents in Network Management

58

Application of Mobile Agents in Network Management

ABSTRACT

According to Case et al., (1990) and Pinheiro et al., (1999), the present day ways of managing network systems as represented by Simple Network Management Protocol (SNMP) are based on the centralized client-server approach. In this approach, the central server (manager) retrieves raw management data from all the clients (network elements) being managed. This raw data is centrally collated and processed by the server for every management cycle. This is good enough for small, reliable and fast networks that are particularly localized in nature. In present-day situation where networks extend across cities, regions, countries, and continents; as well as in mobile applications where connected links are very slow and expensive, this may lead to inefficiency in terms of bandwidth utilization, resource wastage, processing bottleneck on the server, and other problems that could be caused by communication failures between the server and the clients being managed.

The objective of this work is to demonstrate that a more efficient approach to Network Management could be achieved by implementing a distributed system through the use of Mobile Agents (MA). Typically, Biesoczad et al., (1998) and Gavalas et al., (1999) were among the early researchers who investigated the concept and possibilities of using Mobile Agents in the management of network components. Using this concept, a central manager generates software agents and dispatches them with predefined itinerary and specific network management tasks; visiting each network element to be managed; retrieving, analyzing, and only returning computed results back to the central manager from where the agents originated. This way, the central manager does not need to pull in huge volume of raw management data from the many dispersed network elements being managed for central processing. This conserves expensive network bandwidth, relieves the central server of the mundane task of processing raw management data that could result in processing bottleneck, and ensures that only computed results get transmitted to the central server for display to the network administrator.

JAVA Agent Development Environment (JADE) platform was used to develop a Mobile Agent based network management system. Issues covered by this research include Status monitoring of network elements, simple network management protocol table filtering, and Global filtering on a UNICEF (United Nations Children’s Fund) network environment that spanned across five cities (Abuja, Bauchi, Enugu, Kaduna, and Lagos) in Nigeria. MA’s were generated and dispatched by a central server in Abuja, visiting network elements to be managed in Bauchi, Enugu, and Lagos. In each case, volume of data in bytes transferred in both directions as well as transit time in seconds taken by this data were noted. Results were compared with those obtained from the same experiments conducted using SNMP approach. Final analysis shows that while performance was slower using MA as compared to SNMP approach, bandwidth utilization was far much better with MA. Even the slower performance exhibited by MA was mainly due to the JADE platform used during the implementation stage. Other studies have shown that better throughput is attainable using other development platforms. Also covered by this work are some security issues relating to mobile agents in a networked environment.

TABLE OF CONTENTS

DECLARATION ……………………………………………………………………………………………. ii
CERTIFICATION ………………………………………………………………………………………….. iii
ACKNOWLEDGEMNT …………………………………………………………………………………. iv
ABSTRACT …………………………………………………………………………………………………… v
TABLE OF CONTENTS ……………………………………………………………………………….. vii
LIST OF FIGURES AND PLATES ………………………………………………………………….. xi
LIST OF TABLES ………………………………………………………………………………………… xii
ABBREVIATIONS ………………………………………………………………………………………. xiv
DEFINITIONS …………………………………………………………………………………………….. xvi
CHAPTER 1 ………………………………………………………………………………………………….. 1
GENERAL INTRODUCTION ………………………………………………………………………….. 1
1.1 BACKGROUND OF THE STUDY …………………………………………………………. 1
1.2 STATEMENT OF THE PROBLEM ………………………………………………………… 2
1.3 AIMS AND OBJECTIVES OF THE RESEARCH …………………………………….. 3
1.4 RESEARCH QUESTIONS AND HYPOTHESIS ………………………………………. 4
1.5 ORGANIZATION OF THE THESIS ……………………………………………………….. 5
1.6 CONTRIBUTION TO KNOWLEDGE …………………………………………………….. 6
viii
CHAPTER 2 ………………………………………………………………………………………………….. 7
LITERATURE REVIEW …………………………………………………………………………………. 7
2.1 OVERVIEW OF NETWORK MANAGEMENT ……………………………………….. 7
2.1.1 Simple Network Management Protocol …………………………………………………. 8
2.1.2 Management by Delegation (MbD) ………………………………………………………. 9
2.1.3 Remote Monitoring (RMON) …………………………………………………………….. 10
2.1.4 Mobile Agents (MA) ………………………………………………………………………… 11
2.2 JADE – JAVA AGENT DEVELOPMENT FRAMEWORK ………………………. 11
CHAPTER 3 ………………………………………………………………………………………………… 14
MATERIALS AND METHODS ……………………………………………………………………… 14
3.1 SYSTEM ARCHITECTURE ………………………………………………………………… 14
3.1.1 Network Manager ……………………………………………………………………………. 14
3.1.2 Network Element …………………………………………………………………………….. 16
3.2 MANAGER AGENT …………………………………………………………………………… 16
3.2.1 Communication of components ………………………………………………………….. 17
3.2.1.1 Receiving a Register Message ……………………………………………………….. 17
3.2.2 Manager Agent Behaviours ……………………………………………………………….. 19
3.2.2.2 Implementing Agent Bebaviour Classes …………………………………………… 20
3.3 DAEMON AGENT (NEAGENT) ………………………………………………………….. 21
3.4 HEALTH FUNCTION (PERFORMANCE INDICATORS) ………………………. 23
3.4.1 Interface Utilization …………………………………………………………………………. 24
3.4.2 Error Rate and Accuracy …………………………………………………………………… 25
3.4.3 Internet Protocol (IP) Output Datagrams Discard Rate …………………………… 26
3.5 MESSAGE DEFINITION ……………………………………………………………………. 27
3.6 COMMUNICATION OF MANAGER AGENT WITH DAEMON AGENT…. 27
3.7 MOBILE AGENT (MA) ………………………………………………………………………. 29
3.7.1 SNMP table filtering ………………………………………………………………………… 30
3.7.2 Global Filtering ……………………………………………………………………………….. 34
CHAPTER 4 ………………………………………………………………………………………………… 37
RESULTS AND DISCUSSION ………………………………………………………………………. 37
4.1 SECURITY PROBLEM ………………………………………………………………………. 37
4.1.1 Protecting NE from malicious MA ……………………………………………………… 38
4.1.2 Protecting the important data in MA …………………………………………………… 40
4.1.3 Secure the Communication Channel ……………………………………………………. 42
4.2 SCALABILITY ………………………………………………………………………………….. 43
CHAPTER 5 …………………………………………………………………………………………………… 46
SUMMARY, CONCLUSION AND RECOMMENDATION ……………………………….. 46
5.1 SUMMARY ………………………………………………………………………………………. 46
5.2 CONCLUSION ………………………………………………………………………………….. 47
5.3 RECOMMENDATION AND FUTURE WORK ……………………………………… 47
5.4 REFERENCES …………………………………………………………………………………… 49

CHAPTER ONE

GENERAL INTRODUCTION

1.1 BACKGROUND OF THE STUDY

According to Case et al., (1990) and Pinheiro et al., (1999), the present day ways of managing network systems as represented by Simple Network Management Protocol (SNMP) are based on the client-server architecture, where a central station, called the manager collects and analyses raw data retrieved from physically distributed network elements. In those systems, management data are stored in a standard structure maintained on the network elements (NE) to be managed, such as Management Information Base (MIB) Objects Tree in SNMP. Under this setup, there is always a routine/task at each network element, such as SNMPD (Simple Network Management Protocol Daemon) running on Linux, which periodically fetches and returns raw management data in response to requests from the network manager. Network management system based on Client/Server paradigm like this normally requires transferring large amount of management data between the manager and managed network elements. The large amount of data not only requires considerable
bandwidth, but also can cause a processing bottleneck at the manager. This is particularly disturbing considering the fact that the bulk of the transmitted data may end up being irrelevant and hence, discarded by the manager on arrival. As current networks grow larger and more complicated, the problem becomes more severe (Case et al.,
1990).

1.2 STATEMENT OF THE PROBLEM

As highlighted in section 1.1, network management based on Client/Server approach poses a number of challenges due to its centralized nature. Some of these challenges as extensively discussed by Bivens et al., (1999) and Bohoris et al., (2000) are:

a. Bandwidth intensive: For larger networks, management data being transferred between the manager and the remotely located network objects being managed becomes huge and consumes a lot of communication bandwidth. Bandwidth today stands out as one of the most expensive resource in a typical networked environment. This is especially the case when the network is geographically dispersed and leased communication channels have to be utilized.

b. Processing bottleneck at Network Manager: As the data is raw, the network manager spends a lot of time analyzing, hence, sometimes causing processing to be slower than desired. This impact negatively on the entire management domain where similar processing needs to be performed by the same manager for other managed network elements. It becomes more serious when results are needed on real-time basis and the manager is not forthcoming.

c. Resource Wastage: Worse again, in most cases, substantial part of the data is not needed and therefore discarded by the network manager after processing, resulting in overall resource wastage (bandwidth, processing time, storage,
memory, and so forth).

d. Effect of Communication failure: As is the case with all networked environments, sometimes, communication breakdown does occur between the manager and some or all the network objects being managed. This impact negatively on the ability of the network manager to collate management data for processing.

Together, these challenges form the overall motivation for this research work and at the same time, help in defining the objectives set to be achieved by the research as detailed in section 1.3 below.

1.3 AIMS AND OBJECTIVES OF THE RESEARCH

Overall, this project work tries to solve the problems listed in section 1.2 by applying a distributed management mechanism which overcomes the limitations posed by the traditional centralized Client/Server architecture as presented by SNMP. The work demonstrates that mobile agents could be used to perform network management functions more efficiently in terms of bandwidth utilization as compared to the traditional SNMP approach that requires a lot of communication bandwidth which effectively reduces its utilization span in terms of network size. According to Bivens et al.,(1999); Gavalas et al., (2000); Paliafito and Tomarchio (2000); Bohoris et al., (2000); and Griffin et al., (2001); Mobile Agents could be used as a decentralized approach to network management which greatly reduces the workload at the management server location by otherwise delegating network management responsibilities to the network elements being managed. To be able to achieve these desired objectives, the following approaches to network management were extensively reviewed:

a. Remote Monitoring (RMON);

b. Management by Delegation (MbD); and

c. Mobile Agents.

Each of these three (3) alternatives mentioned above introduces some degree of decentralization. However, the third solution, the use of Mobile Agents (MA) technology to distribute and delegate management tasks has been used to implement the following aspects of network management:

a. Network elements’ status monitoring,

b. SNMP table filtering, and

c. Global filtering.

1.4 RESEARCH QUESTIONS AND HYPOTHESIS

Distinguished scholars such as Bivens et al.,(1999); Gavalas et al., (2000); Paliafito and Tomarchio (2000); Bohoris et al., (2000); Griffin et al., (2001) have extensively investigated the use of Mobile Agents (MA) technology to distribute and delegate management tasks in the past.

Additionally, mobile Agent frameworks have already attracted a lot of attention in recent years. A lot of research is currently being carried out to assess the applicability of agent technology to network management and control environment. It has been argued by Gavalas et al., (2000); Paliafito and Tomarchio (2000); Rubinstein et al., (2002) that MAs have some superior features over SNMP, MbD and RMON. There is a general agreement that MA can be used to alleviate the network manager workload and reduce the bandwidth usage by delegation of authority from the manager to MA. Those advancing this argument believe that MA is more flexible and could be instantly customized by user’s requirement and launched from the manager. It can visit each network element according to the itinerary table, computing and compressing the management data locally, only returning the result to the network manager. By moving a portion of the “intelligence” to the nodes where data are resident, many of the management decisions could be taken locally, thus avoiding the transfer of large amounts of data from the remote nodes to the central manager (Paliafito and Tomarchio 2000).

Applicability of Mobile Agents in Network management was implemented using JAVA Agent Development Environment (JADE) platform. Applications implemented in this system are:

a. Network element’s status monitoring,

b. SNMP table filtering, and

c. Global filtering.

This thesis described the architecture, design and implementation details of these applications. Common security problems of mobile agent were also investigated.

Security is a crucial problem to the feasibility of mobile agent, especially to the network management domain which has great security requirements. Finally, the performance of the system was tested and potential problems analyzed.

1.5 ORGANIZATION OF THE THESIS

An introduction to some background knowledge related to this research is given in chapter 2. This includes brief overview of JADE, SNMP; a brief description of Remote Monitoring (RMON), Management by delegation (MbD) and Mobile Agents (MA).

The overall overview of the system’s architecture is given in Chapter 3. Also discussed in this chapter are the detailed design and implementation issues.

Results, discussions, as well as identified related problems, such as securing the whole system and scalability test are explained in Chapter 4.

Chapter 5 summarized, concluded and highlighted possible future work on this research area.

1.6 CONTRIBUTION TO KNOWLEDGE

The thesis contributes to the area of network management by demonstrating that an entirely new emerging concept of Mobile Agents technology could be used to delegate and distribute network management responsibilities across network objects; as opposed to the traditional Clients/Server central approach which though simple, poses a number of operational challenges as regards to bandwidth utilization, processing bottleneck at manager location, and difficulty in handling single point of failure. Although, an entirely new area in modern computing, a number of studies have been undertaken by others in the area of Mobile Agents application in network management. Bieszczad et al., (1998) described theoretical views on applications of MA for network management.

They discussed potential uses of MA’s in network management and defined software agents and a navigation model that determines agents’ mobility. They further listed potential advantages and disadvantages MA’s. Gavalas et al., (1999) presented the application of MA in bulk transfers of network monitoring data, data aggregation and acquiring atomic SNMP table views. They analyzed the usage of mobile agents in network management with regards to the bandwidth utilization. Pinheiro et al., (1999) described a conceptual model which collects management related data across a changing set of networked components and periodically computes aggregated statistics using MA. This thesis draws ideas from above concepts and applies them differently in the area of MA application to network management; with particular interest in Network elements’ status monitoring, SNMP table filtering, and global filtering.

Get Full Work


Disclaimer:
Using this Service/Resources:
You are allowed to use the original model papers you will receive in the following ways:
1. As a source for additional understanding of the subject.
2. As a source for ideas for your own research work (if properly referenced).
3. For PROPER paraphrasing (see your university definition of plagiarism and acceptable paraphrase)
4. Direct citing (if referenced properly)
Thank you so much for your respect to the authors copyright.