using mobile agents to improve performance of network management ...

USING MOBILE AGENTS TO IMPROVE PERFORMANCE OF NETWORK MANAGEMENT OPERATIONS Iwan Adhicandra, Colin Pattinson, Ebrahim Shaghouei Computer Communications Research Group, School of Computing, Leeds Metropolitan University, Beckett Park, Leeds LS6 3QS, United Kingdom {i.adhicandra; c.pattinson; e.shaghouei}@lmu.ac.uk Abstract - Currently, most network management systems operate SNMP (Simple Network Management Protocol). These protocols use the client-server model, on which the management station acts as a client that provides a user interface to the network manager and interacts with agents, which are servers that manage remote access to the Management Information Base (MIB). In certain circumstances (e.g. at times of network stress), this clientserver interaction generates significant traffic that overloads the management station. A distributed paradigm is a revision way to perform management functions when networks grow significantly. In this sense, mobile agent is an option to distribute the network management. These agents move to the place where data are stored and select information the user wants. They decentralise processing and control, and, as a consequence, reduce the traffic around the management station, and distribute processing load. The purpose of this study is to investigate the effect of using a mobile agent as compared to a static agent in alleviating loading created by network management protocols and to investigate the performance behaviour of network management concepts using simulation.

I. INTRODUCTION Network management fundamentally involves monitoring and controlling the devices connected in a network by collecting and analyzing data from the devices [1]. It gives network administrators the flexibility of managing the whole network from a single place. SNMP is often used in centralized network management environments. It is the dominant protocol for network management and supports the operations: Get-Request; Get Next-Request Get-Response, Set-Request and Trap. In SNMP a management application uses the manager protocol to communicate with the managed system, which uses the agent protocol to communicate with the MIB and the manager protocol. Processing of managed data is done at the management station. Network management stations interact with SNMP agents in managed nodes. Each SNMP agent is essentially a daemon process that responds to requests from management stations. Its drawbacks include information bottleneck at the manager, lack of scalability, excessive processing load at manager, heavy usage of network bandwidth by network management actions, management intelligence too centralized. An alternative is distributed network management, in which the centralized management strategy is replaced by interoperable management systems. Distributed management solves the problems with centralized management to some

ISBN: 1-9025-6009-4 © 2003 PGNet

extent. However, it still has some drawbacks like limited scalability and complex coordination mechanisms between management stations. One important functional area of network management is Performance management, which involves gathering statistics about network traffic and schemes to condense and present data. Measuring performance of networks using centralized SNMP based management is very difficult due to reasons like network delays and information bottleneck at the central management station. The latest trend is to deploy mobile agents to manage today’s large heterogeneous networks. Mobile agents are special software objects that are autonomous and have the ability to migrate from one node to another node, carrying logic and data, performing actions on behalf of the user. Mobile agent based network management is to equip agents with network management capabilities and allow them to issue requests to managed devices (or nodes) after migrating to these nodes. Mobile agents give the flexibility of analyzing the managed node locally. Instead of querying the managed node for every fixed interval and analyzing the performance from management station, mobile agent can be dispatched to analyze the node locally. The purpose of this study is to investigate the effect of using a mobile agent as compared to a static agent in alleviating loading created by network management protocols and to investigate the performance behaviour of network management concepts using simulation [20]. This paper is organized in the following way. In Section 2, we review literature in mobile agent implementation and performance issues in network management. In section 3, we describe a proposed architecture for mobile agent in network management. In section 4, we explain the performance model for network management using Health Function. Section 5 presents a proposed method in mobile agent implementation using simulation. We conclude the paper in Section 6 with remarks on future work. II.

MOBILE AGENTS IMPLEMENTATION NETWORK MANAGEMENT

IN

In past few years, many researchers have investigated mobile agent implementation in network management. One important example is the PMP Project effort [2] wherein mobile code was used for managing the

network. Network manager is a suite of simple tools that interact with agents located on nodes connected in a network, as well as other tools used for managing networks. Each of the tools can be used for actual management of networks. Less stress was laid on embedding existing SNMP into a mobile code framework. This results in incompatibility with SNMP based systems. Bieszczad et al. [3] described theoretical views on application of mobile agents for network management. Gavalas et al. [4] presented the application of mobile agents 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 regard to the bandwidth utilization. Pinheiro et al, [5] described a conceptual model which collects management related data across a changing set of networked components and periodically computes aggregated statistics using mobile agents. The investigation of the performance of mobile agents in network management is recent. Baldi et al. [6] estimate the tradeoffs of mobile code design concepts in network management applications by developing a quantitative model that provides the bandwidth used by traditional and mobile code design of management functionalities. Bohoris et al. [7] introduce a performance comparison of mobile agents, CORBA, and Java-RMI by using one network element on a ATM network. They use an array of objects which are not real data in order to gain the response time and bandwidth utilization. Gavalas et al. [8] investigate bandwidth utilization for SNMP and mobile agent. The paper provides an experimental implementation, which is presented in terms of bandwidth consumption and response time to gain an aggregation of multiple variables on a LAN of a few nodes. The measurements of bandwidth utilisation of mobile agent and client-server applications is carried out by [9] on an Ethernet LAN of a few nodes. They also present a single case comparison of response time for the mobile agent and the client-server. Kona and Xu [10] provide a performance method to use SNMP and mobile agent and also proposed a framework for mobile agent architecture. They built their experiment using Naplet framework which uses Java programming language. Rubinstein and Duarte [11] simulate management tasks performed by mobile agents and SNMP ones, comparing both approaches, on a topology that consists of a LAN of managed network elements connected to the management station by a bottleneck link. They present quantitavive measurement using NS simulation tool particularly on network management performance using static and mobile agent. III. MOBILE AGENT ARCHITECTURE Mobile agents perform their tasks by moving between different networked computers. A mobile agent can migrate from one machine to another, carrying data about its state that includes information obtained from previous task executions. As the number of visited nodes grows,

the mobile agent size also increases, making migration harder. One possible solution to this problem is to visit a fixed number of nodes, return to the agent home or send all data to it (reducing the mobile agent size), and start the task again on the remaining nodes. The initial size of a mobile agent also affects agent performance since the larger the size, the more difficult the migration. This size depends on the task to be carried out and on the language used to implement it. In SNMP, network management does not scale when the size or the complexity of the network increases because of centralized-processing and control. Mobile agents can be used in order to solve this problem and it is important to investigate when they improve the management efficiency To take advantage of mobile agents for network management, we propose a flexible architecture, which forms a layer over conventional SNMP based management. This ensures that the advantages of SNMP are not lost and also serves the purpose of managing legacy SNMP based systems.

MS

NC

SA

SA

MIB

MAS MAS MA MA

NC MAS MS=Management Station NC=Network Component SA= Static Agent MA= Mobile Agent MAS= MA System

MA

Figure 1. Proposed mobile agents architecture for client-server connection.

Figure 1 shows the proposed architecture for network management using mobile agents. The manager is given the flexibility of deciding whether to use SNMP or mobile agents. In this approach the station assumes responsibilities of a client. All managed nodes are

servers, which have mobile agent environment and respond to SNMP queries from mobile agents when they visit the servers and manipulate data locally. When the client needs access to data on a network connected device, it does not talk directly to the server over the network. Instead, the client actually dispatches a mobile agent to the server’s machine. Once on the server’s machine, the MA makes its requests to the server directly. When the entire transaction is complete, the mobile agent returns to the management station with the results. The architecture provides Java-compliant interfaces to network management services. We propose to use Aglets Software Developer Kit (ASDK) [12] as the agent development environment. It provides a modular structure, easy-to-use API for programming of mobile agents and excellent documentation. To interact with the SNMP agent we propose to use AdventNet SNMP[13]. AdventNet SNMP provides a set of Java tools for creating cross platform Java and Webbased SNMP network management applications. This package provides a set of classes, which could be used to facilitate communication between a managed device (a device with an SNMP agent, e.g. a router), and an SNMP manager or management application. Using the mobile agent paradigm can bring some interesting advantages when compared to traditional client/server solutions [14]. It can reduce network traffic; it can provide more scalability; it allows the use of disconnected computing; and it provides more flexibility in the development and maintenance of the applications. IV. ANALYTICAL APPROACH In order to investigate the performance evaluation using mobile agent, we need to do some analytical solutions as following: IV.1. Transmit Model This model could be used to monitor the performance of a set of managed nodes over a particular interval of time. It can use Health Functions, which indicates system state or efficiency of the node and could be viewed as a way to compress management data and evaluate the performance of any element. A performance management application can use ifInOctets and ifOutOctets of the interfaces group in MIB to compute the percentage utilization of an interface over an interval of time. To perform this computation, two different polling intervals are required: one to find total bytes per second at time x and another to find total bytes per second at time y. The following equation computes utilization, U(t) for the polling interval (x-y) seconds [15]: U(t) = [((ifInOctetsy-ifInOctetsx)+(ifOutOctetsy–ifOutOctetsy))*8] / (y-x)*ifSpeed

where, ifspeed is the bandwidth of the interface, inOctetsx is the bytes received by the interface at time x,

IfOutOctetsx is the bytes sent by the interface at time x, (y-x) is the polling interval. By using transmit model we could dispatch MA to each of the managed nodes and MA could calculate the utilization for every polling interval over an extended period of time. Here the MA manipulates the data locally at the managed node. After the time period for which the analysis is required, which is equal to the sum of all the polling intervals, the MA returns to the management station and the reduced data set could be displayed in a graph. IV.2. Route Model In this model a mobile agent visits the set of nodes to be managed sequentially. The mobile agent is configured with the list of nodes to be visited during its itinerary and also the SNMP statistics to be analyzed. Health function is also used to calculate percentages of input and output errors on an interface. It is a cumulative factor of 8 MIB variables [15]: Percent input errors = ((ifInErrors)/(total packetsreceived))*100 Percent output errors = ((ifOutErrors)/(total packets sent))*100

where, total packets received = (ifInUcastPkts+ifInBroadcasts+ifInMulticasts) total packets sent = (ifOutUcastPkts+ifOutBroadcasts+ifOutMulticasts)

If the interface error rate is more than 1%, then there is a problem with the interface of the machine. If the error rate is less than 1% and network shows poor performance, then it could be deduced that there is a problem with the media. At each managed node MA interacts with the SNMP Agent, calculates percentage input and output errors locally on the managed node. After visiting all the nodes it returns back to the manager. V. SIMULATION APPROACH The behaviour of mobile agents in carrying out network management tasks is assessed by simulation studies. The OPNET Modeler is used [16]. This discrete-event simulator provides protocols and mechanisms to simulate networks with node and link abstractions. We propose to use multi tier application (MTA) topologies, which are similar in shape to client server application. The scenarios consist of one client with different number of servers or managed nodes. These are chosen to show the effect of increased managed node for network performance. A fair evaluation can only be achieved with careful selection of simulation parameters. Table 1 summarises important simulation parameters used in these experiments. These parameters are those required by OPNET package to determine an application level configuration. These are request information, response

information and session information. These are chosen in order to see effect of adopting mobile agent in the network on the overall network performance. A “Response” is taken to be the actions required to complete a full task. It starts from the time when the manager sends the request to the mobile agent and end when it receives the response back. A “Request” is the information sent from the manager to the mobile agent, and a “Response” is that from the mobile agent to the manager. Table 1. Parameters used in simulation Grouping Request Information Response Information Session Information

Parameters Generation Rate PDF Packet Size (Kbytes) Packet Size PDF Packet Size (Kbytes) Packet Size PDF Background Utilization (%) Time (sec)

Value Exponential 5 Poisson 5.050 Poisson 25 100

The simulation model assumes that links and nodes have no load, and links are error-free. User Datagram Protocol (UDP) is used in all simulations. The initial size of the mobile agent is chosen as 5 kbytes [17][18]. Each request or response of a variable is 0.050 kbytes long.

Corresponding with research conducted by [17] and [18], we choose request generation delay and response generation delay values, of 0.1 sec and 0.5 sec respectively. The request generation rates PDF are chosen to be exponential, whereas request packet size PDF and response packet size PDF are chosen to be exponential [19]. SNMP sends requests to all elements to be managed (one after receiving the response from the other); the mobile agent goes to an element to be managed, gathers the variable, and visits all other elements. After finishing, the mobile agent returns to the management station. Initial result on response time received, network operations using static agent show better result on the condition of small number of nodes (in this case 4 nodes). In terms of traffic at client side and server side, initial result shows that there is a reduction of traffic received when the number of nodes were increased and mobile agent performance shows better result. In terms of load on the managed node, there is a reduction of load received when the number of nodes were increased and again mobile agents performance shows better results than static agent. VI. CONCLUSIONS AND FUTURE WORK This paper has presented a method to measure the performance of network management operations using client server and mobile agent approach. The initial results show that mobile agent perform better when the number of managed nodes increase significantly. However, this initial result still needs to be improved to give a better result particularly on response time using mobile agents and network operations using large number of managed nodes. This research work is currently in progress and we plan to extend the model with more extensions for client and server model, to investigate the use of process model in each nodes, investigation of other environments such as wireless network, and to conduct more some experimental work in order to find an accurate representative model. REFERENCES

Figure 2. Initial results using simulation

[1] William Stallings. SNMP, SNMPv2 and RMON: Practical Network Management. Addison-Wesley 1999. [2] Perpetuum Mobile Procura (PMP) Project, http://www.sce.carleton.ca/netmanage/ [3] Andrzej Bieszczad, Bernard Pagurek and Tony White. Mobile Agents for Network Management. IEEE Communications survey-98. [4] D. Gavalas, D. Greenwood, M. Ghanbari and M.O. Mahogany. Advance network monitoring applications based on mobile intelligent agent technology. [5] Robert Pinheiro, Alex Poylisher,Hamish Caldwell. Mobile Agents for Aggregation of Network Management Data. Telecordia Technologies.

[6] Baldi, M. and G.P. Picco, Evaluating the Tradeoffs of Mobile Code Design Paradigms in Network Management Applications, 20th International Conference on Software Engineering (ICSE’98), Kyoto, Japan, Apr. 1998, pp146-155. [7] Bohoris, C. Pavlou, G. and H. Cruickshank, Using Mobile Agents for Network Performance Management, IEEE/IFIP Network Operations and Management Symposium, Honolulu, Hawaii, 2000. [8] Gavalas, D., Greenwood, D., Ghanbari, M., and M. O’Mahony, Using Mobile Agents for Distributed Network Performance Management, the 3rd International Workshop on Intelligent Agents for Telecommunication Applications (IATA’99), Stockholm, Sweden, Aug. 1999. [9] Sahai, A and C. Morin, Towards Distributed and Dynamic Network Management, IEEE/IFIP Network Operations and Management Symposium (NOMS’98), New Orleans, USA, Feb 1998. [10] Kona, M.K., and C.Z. Xu, A Framework for Network Management using Mobile Agents, in proceeding of ICEC 2001 [11] Rubinstein, M and O. C. M. B. Duarte (1999), Evaluating Tradeoffs of Mobile Agents in Network Management, Networking and Information Systems Journal 2, pp.237-252 Feb. 1999. [12] Aglets Software Development Kit (ASDK), http://www.trl.ibm.com/aglets/ [13] AdventNet, http://www.adventnet.com/ [14] General Magic. Odyssey web page. http://www.genmagic.com/agents/odyssey.html

[15] Allan Lienwand, K. Fang Conroy. Network Management: A Practical Perspective. Addison Wesley 1996. [16] OPNET Documentation, OPNET Technologies, Inc. [Internet] http://www.opnet.com [17] Rubinstein, M, O. C. M. B. Duarte, and G. Pujolle, Evaluating the Network Performance Management Based on Mobile Agents, in proceeding of 2nd International Workshop of Mobile Agents for Telecommunication Applications (MATA), Paris, France, September 2000. [18] Puliafito, A., and O. Tomarchio, Using Mobile Agents to Implement Flexible Network Management Strategies, Computer Communications, 23, 8, pp.708-719, April 2000. [19] Pattinson, C., A Study of the Behaviour of the Simple Network Management Protocol, in proceeding of 12th International Workshop on Distributed Systems: Operations and Management (DSOM), Nancy, France, October, 2001 [20] Adhicandra, I., and C. Pattinson, Performance Evaluation of Network Management Operations, in proceeding of 3rd Annual Symposium of Postgraduate Networking Conference (PGNET), Liverpool, UK, pp. 210-214, June 2002 [21] Indrawan, M., S. Krishnaswamy, and T Ranjan, Using Mobile Agents to Support Unreliable Database Retrieval Operations, in proceeding of the 17th International Conference on Advanced Information Networking and Applications (AINA’03)