design and implementation of a web-based internet performance

INTERNATIONAL JOURNAL OF NETWORK MANAGEMENTInt. J. Network Mgmt. 9, 309–321 (1999)

Design and Implementation of a Web-based InternetPerformance Management System Using SNMP MIB-II

By Seong Jin Ahn,Ł Seung Keun Yoo and Jin Wook Chung

This article is aimed at defining items of analysis using SNMP MIB-II forthe purpose of analyzing the performance of Internet-based networksrunning on TCP/IP protocol, and then utilizing these items, in conjunctionwith various Web technology and JAVA, to design and implement aWeb-based interface of a management system to analyze the performanceof the Internet. Copyright 1999 John Wiley & Sons, Ltd.

Introduction

W ith the rapid development ofinformation technology, TCP/IP-based Internet is now being used

widely in various fields, providing the end userwith a basic structure for distributed servicearound the globe. Such an evolution of Internet,coupled with the increasing traffic associated withan ever broader user base, is seriously impairingnetwork performance and is calling for a need for acomplete redesign of the system. It is therefore verytimely to ensure a good performance managementof the Internet to sustain a stable network.1,2

Performance management allows the systemmanager to maintain an optimum system per-formance by collecting, analyzing and making adecision on, the traffic of the overall network,sub-network, and particular nodes. Furthermore, itprovides the user with vital information regardingfuture expansion or modification of the system. Forthe maintenance of communication equipment inthe Internet, SNMP (Simple Network Management

Protocol) is used to exchange system managementinformation between the manager system and themanaged system.

Moreover, in today’s complex network environ-ments, the manager must bear the inconvenience ofusing a combination of tools of at least two differenttypes and of course only those which are supportedin the system or network he/she is managing. Anadditional problem is the limitations of scalabil-ity of the currently used centralized managementtools, which make it very hard to handle very largenetworks. In order to solve these problems, oneneeds to utilize Web-related technology or otherInternet-based technology such as JAVA to createa system management application which can effec-tively overcome the limits of existing managementtools as technology.3,4

For this paper, items of analysis that are ofinterest were defined and sorted from the analysisparameters for Internet performance managementwhich in turn were extracted from SNMP MIB-II.Based on these items was then formulated, withthe help of Web-related technology and JAVA,

Seong Jin Ahn received his Ph D from the Sung Kyun Kwan University in 1998. His research interests include Web-based network management,network design and traffic forecasting.

Seung Keun Yoo graduated from Sung Kyun Kwan University, Korea, in 1998. He is currently a MS candidate in the same university.

Jin Wook Chung received his Ph D from the Seoul National University in 1991. He is a Professor in the School of Electrical and ComputerEngineering, Sung Kyun Kwan University, Korea.

ŁCorrespondence to: Seong Jin Ahn, Information and Communication Laboratory, School of Electrical and Computer Engineering, Sung Kyun

Kwan University, 300 Chunchun-dong Jangan-gu, Suwon City, South Korea.Email: [email protected]

Copyright 1999 John Wiley & Sons, Ltd. CCC 1055–7148/99/050309–13$17.50

310 SEONG JIN AHN ET AL.

a visualization system, that is, a Web interfaceof the Internet management system, carrying outthe actual management task. Moreover, as theinterface is based on a client/server structure,a protocol by the name of MATP (ManagementApplication Transfer Protocol) has been definedfor the purpose of exchanging messages betweenthe client and the server and between the serverand the analysis system executing accumulativeanalysis. As a result, it was possible to conceivewith the help of JAVA and other Web-relatedtechnologies, a network management scheme thatis completely platform independent. A systematicanalysis of the collected management informationand presentation of the results in the form of a

graph allow the user to easily access and use thedata obtained.5,6

Related research to this paper includes thedevelopment of a private network/intranet trafficanalysis system using Web technology and RMONMIB and a Web-based service application analysissystem aimed at analyzing TCP/IP traffic with theobjective to study the types and the frequency ofTCP/IP network services.7

Internet PerformanceManagement

Performance management seeks to inspect theactions of the managed system and increase the

Analysis type Analysis circle Analysis item

Real-timeanalysis

Internet performanceanalysis

Link utilization, availability, interface packettransmit/receive rate, I/O traffic rate,broadcast traffic rate, system packet I/Orate, packet forwarding rate

Internet fault analysis Error receive rate, interface packet errorrate, system packet error rate, systemmemory load rate, packet forwarding errorrate, routing error rate

Management trafficanalysis

Management traffic utilization

Elementaryanalysis

Internet performanceanalysis

Link utilization, availability, system downcount,

Detailed analysis Internet synthesisanalysis

Synthesis analysis, link analysis, deviceanalysis, device application analysis

Internet timely analysis,monthly analysis, weeklyanalysis, daily analysis

Error receive rate, interface packettransmit/receive rate, interface packet errorrate, input traffic analysis, output trafficanalysis, broadcast traffic analysis, systemmemory load rate, packet forwarding rate,routing error rate

Internet comparativeanalysis

Inbound/outbound byte,inbound/outbound packet, broadcasttraffic analysis (broadcast/unicast), byte perpacket (transmit/receive), normal/abnormalpacket analysis, system packet error rate,system packet error I/O analysis

Internet termcomparative analysis

Inbound/outbound byte,inbound/outbound packet, broadcasttraffic analysis (broadcast/unicast), byte perpacket (transmit/receive), normal/abnormalpacket analysis, system packet error rate,system packet error I/O analysis

Table 1. Internet performance analysis items

Copyright 1999 John Wiley & Sons, Ltd. Int. J. Network Mgmt. 9, 309–321 (1999)

A WEB-BASED INTERNET PERFORMANCE MANAGEMENT SYSTEM 311

efficiency of communications. It involves collectingdata on such performance indicators as utilizationand collision rate for the overall network, subnet-work, specific segments and nodes. The collecteddata are then analyzed to determine the currentperformance of the network and are used to main-tain an optimum performance. These data alsoprovide vital information during system expansionor modification.

The items of analysis for the performance man-agement of the Internet are formulated using thediverse performance analysis parameters extractedfrom SNMP MIB-II. These items are classifiedand grouped according to the type of analysis.8

Real-time analysis is used when one wishes toanalyze the performance of the Internet-based net-work equipment at the moment of request. Bycontrast, elementary and detailed analyses involvethe analysis of data collected over a given periodof time. Elementary analysis deals with measuringelementary indicators for the analysis of Internetperformance such as utilization and availabilityover a period specified by the user. Detailed analy-sis is used to obtain a comprehensive and detailedanalysis of the performance of the Internet-basedmanaged system over a given period of time.9

Table 1 is an explanation of the MIB objectanalysis items concerned with the above analysisitems.

Design and Realization of theWeb Interface

—Overview of the Design—

The Internet performance management systemselects those managed systems for which the per-formance on the Internet of the network equipmentwill be managed. Information on these selectedsystems will then be collected and analyzed. Theresults will be used to ensure a stable networkby enabling a smooth traffic flow. The systemhas a client/server structure with the client beingresponsible for the user interface and the reportingof the results and the server being responsible forthe real-time analysis of each analysis item andsending the results to the client. Depending on themanagement objective, an accumulated analysissystem can be used whereby data collected overa specified duration can be utilized to carry out a

performance analysis of the equipment over thatvery duration.

The Internet performance management systemmentioned in this paper specifically means theWeb-based graphic user interface with the assis-tance of which a distant user connects himself to theInternet management system via a web browserand carries out a series of activities regardingthe performance of the concerned network equip-ment. Figure 1 is a structural diagram of the entireInternet performance management system whichconsists of a visualization client functioning asclient; a visualization server functioning as server;and an accumulated analysis system (SPES: SNMP-based Performance Parameter Extraction System)analyzing the performance of the system over aspecified period of time.

—Visualization Client—

Structure— The visualization client operatesin a remote web browser and serves as userinterface as well as a system that responds touser demands. It is composed of a module thathas for its function the establishment of connectionwith the server, a message processing module, areal-time processing module, a collection demandmodule, an accumulation analysis module, a graphprocessor and a report processor. The function ofeach module of the visualization client system isas follows as illustrated in Figure 1.

The connection management module has for itsfunction the establishment of connection with thevisualization server system as well as transmissionof messages containing the user’s requests to theserver and receiving the server’s response mes-sages. These messages are then processed by themessage processing module which creates the mes-sages requested by the user, analyzes the responsemessages from the visualization server and trans-mits them to the relevant processing modules. Thereal-time processing module is concerned withprocessing the user’s request to analyze the currentperformance of the network equipment. It is thusa user interface that accepts the user’s request:it extracts the necessary data from the server’sresponse to produce a graph; and receives therequest from the user to terminate data output orto disconnect from the visualization server. The



Figure 1. Structural diagram of the entire Internet performance management system

Figure 2. Overall structure of the visualization client system

collection demand module is an user interface thathas the role, of requesting of data collection to theaccumulation analysis system for the purpose ofcarrying out an analysis with the accumulated dataobtained by polling specific network equipment

over a specified amount of time. The module isalso responsible for sending necessary informationto the visualization server and browsing throughthe material whose collection has been requested.The accumulation analysis method serves as a



user interface for elementary and detailed analysisof network equipment over a specified durationon information that has been accumulated by thecollection demand module. The module displaysthe results in the form of a graph or a tabledepending on the preferred type of analysis. Thegraph processor is called for a graph-based reportof results. The graph processor generates graphsaccording to items of analysis and consists of sub-modules responsible for generating line, pie or bargraphs depending on the item of analysis and theuser’s request. The report processor transforms theoutput graphs generated upon the request for eachanalysis into a GIF image and accordingly creates aweb page for report display purposes. The proces-sor also handles the task of calling the response ofthe accumulation analysis module from the server.

Basic operations— From a functional perspec-tive, the visualization client system for Internetperformance management provides the user withfour different services. Specifically, they are: thefunction of real-time analysis for, as the nameimplies, the real-time analysis of a network per-formance; the two functions of elementary anddetailed analyses for the performance analysis ofthe network over an extended period of time; and

the function of collection demand for the collec-tion of the required data. The visualization clientsystem provides the user with several interfacesfor the above services. To use the desired service,the user inputs information such as IP address andcommunity name to the corresponding interfaceand requests the desired service. The part respon-sible for the communication with the visualizationserver creates a message based on the informa-tion provided by the user. This message is thentransmitted via the transmission module to thevisualization server. Finally a reception is made inthe form of a message or data as a response from thevisualization server. If an error occurs while creat-ing a message or during transmission to and fromthe visualization server, the user is notified of theerror and is asked to input new information. Theanalysis result data obtained through the connec-tion with the visualization server are then shown tothe user in the form of either a graph or an HTMLdocument depending on the type of analysis item.The result is displayed on-screen after the datais processed in the relevant processors mentionedabove. The user service for performance analy-sis is made available by the above processes. Thebasic operations of the visualization client systemis illustrated in Figure 3.

Figure 3. Event transition diagram for the visualization client



—The Visualization Server—

Structure— The visualization server processesthe request from the client and the related tasks of:establishing connection, creating messages; han-dling and transmitting requested data; process-ing by item of analysis; establishing connectionwith and transmitting requested information tothe accumulation analysis system; image creation;creation of Web pages for analysis results; andmanagement of requested information. Figure 4 isan illustration of the overall structure of the visu-alization server. The functions of the constituentmodules are as follows.

The SNMP manager carries out the specific func-tion of the SNMP manager system which pollsrelated MIB information for the purpose of theserver’s extraction of analysis information uponthe request message for a real-time analysis. Theconnection management module is in charge ofestablishing connection with the client as wellas receiving demand messages from the client.The module also processes and manages necessaryconnection to send reply messages. Furthermore,it has also the function of transmitting messages

requesting collection of data and accumulationanalysis to the accumulation analysis system. Themessage processing module has the function ofprocessing received messages according to the for-mat requested by the system administrator. Themodule also analyzes the message before sendingto the relevant processor. The analysis module isin charge of real-time processing of each and everyuser service and consists of modules designed tohandle four types of user services. The real-timeanalysis module has for its function the real-timeprocessing of request by the user for analysis ofthe current state of the system. In order to extractvalues for the item of analysis requested fromthe client, the MIB information data are polledby calling the SNMP manager system. The dataobtained in this fashion are then sent to the anal-ysis item processor with the result transmitted tothe client at each polling. The collection demandmodule processes the request for data collectionfor the analysis of the accumulated data thatare polled over a specified duration for a givennetwork’s resources. The module is also responsi-ble for transmitting data requested by the systemadministrator to the accumulation analysis system.

Figure 4. Overall structure of the visualization server



The module also receives replies from the accu-mulation analysis system regarding the result ofset-up following the request for data collection.The accumulation analysis module’s tasks can bedivided into elementary and detailed analyses. Themodule carries out the function of processing therequest for elementary and detailed analyses ofnetwork resources set up over a specified amountof time for information accumulated by the col-lection demand module. Analysis of accumulatedinformation is relegated to the accumulation anal-ysis system by sending a request message. Afterreceiving the results of analysis, the module reportsto the client. The report module’s task is to tabulatethe graph format results analyzed by the managerat the client into a Web page. The analysis itemprocessor extracts analysis results from the man-agement information polled according to item ofanalysis by the real-time processing module. TheSPES manager is a module responsible for the pro-cessing of requests made by the data collectionmodule and the accumulation analysis module.It accomplishes this task through communicationwith the accumulation analysis system and alsohandles such functions as the management ofconnection with the accumulation analysis sys-tem, creation of request messages, management ofreceived messages, data buffering and flow controlon top of the function of sending the data analyzedby the accumulation analysis system to the request-ing modules. The image generator generates a fileout of the GIF image received from the client andtransmits the file information to the Web generator,The Web page generator presents the results of theelementary analysis, in a tabular form, on a Webpage. It is also responsible for the generation ofWeb pages tailored to the type of the analysis itemin real-time and detailed analyses. It also respondsto requests made by the graph processor. The man-agement information controller has for its functionthe generation of HTML and GIF files as well asthe recording, deleting and viewing of these files.

Basic operations— The visualization serverconsists of three subparts which are responsiblefor the transmission of messages with the visual-ization client, the transmission of messages withthe accumulation analysis system and the process-ing of various user services respectively. Uponreceiving a message from the visualization client,the visualization server analyzes the message

and determines to which user service it belongsbefore it transmits the request information tothe relevant service processor. Each service pro-cessing part receives the pertaining information,and executes the processes of real-time analysis,collection demand, elementary analysis, detailedanalysis and reporting. The results are sent to theconnection-related module which in turn createsa message directed to the client. In real-time anal-ysis, SNMP is used to poll performance analysisinformation of the equipment whose analysis isrequested by the user. The performance informa-tion is derived from this analysis information and issent to the visualization client. The report processorreceives from the visualization client, informationand image data for the report and creates an imagefile and a report file. Information pertaining to filemanagement is then stored. The created files aresent to the visualization client via HTTP and aredisplayed to the user in the form of a Web page. Forthe processing of collection demand, the messagereceived from the visualization client is transmit-ted to the accumulation analysis system where acollection is requested. The result is transferredback to the visualization client. Elementary anal-ysis processing is carried out by transmitting therequest message from the visualization client to theaccumulation analysis system. The analysis resultsreceived thereafter are sent to the visualizationclient in the form of an HTML file. The processingof detailed analysis is similar to that of elementaryanalysis; the request message is sent to the clientafter which the analysis results obtained are pre-sented in either a HTML or a graph format accord-ing to the item of analysis. When a HTML format isdesired, a HTML file is created and then transmit-ted. When a graph is requested, on the other hand,the analysis data is transmitted instead in order toallow the visualization client to plot a graph. TheHTML file or the image file created by the visual-ization server are managed by the file managementpart. The basic operations of the visualizationserver are schematically shown in Figure 5.

—Communication between the Clientand Server—

To analyze network traffic, the visualization sys-tem is composed of several subsystems which arenamely, the client, the server and the accumulation



Figure 5. Event transition chart of the visualization server

analysis system. However, as the client runs onthe system along with the web browser, it firsthas to communicate with the server to analyzethe network resources the administrator wishes.Also, in carrying out the tasks of collection oraccumulation analysis, the server communicateswith the accumulation analysis system. In otherwords, it is necessary to establish a protocolfor message exchanges among the server, theclient and the accumulation analysis system. Nat-urally, such a protocol was designed and wascalled MATP (Management Application TransferProtocol).10

Method of communication— When anapplet is loaded onto the administrator’s Webbrowser, the visualization client requests a TCP/IPconnection to the server and sends the requestmessages of the user. The visualization client caninclude data and in this case, the server confirmswith an acknowledgement. Following the user’srequest, the visualization server makes a TCP/IPconnection with the accumulation analysis systemto which it will subsequently send a requestmessage. The accumulation analysis system replieswith a result message. Data can be includedin the message in which case the visualizationsystem replies with an acknowledgement. The

visualization server transmits the results ofthe user’s request processed in this mannerand the corresponding analysis results to thevisualization client. In return, upon reception ofdata from the visualization server, the visualizationclient confirms with an acknowledgement. Afterreceiving all data and completing the output ofresults, the visualization client sends a request anabort message to the server. Figure 6 illustratesthe above-mentioned mechanism of the MATPProtocol.

Definition of a message— Analysis type isa field representing the protocol currently beingused to differentiate it from a possible expansionof the protocol in the future. Message type is a fieldthat sets apart the service currently requested bythe user to the visualization system. Subtype is atype to handle the details of each request in themessage type. Start time and end time are fieldsthat sets the duration of the collection period. Theyhave to be inputted by the user and are furthersubdivided into fields representing the year, thedate, the hour and the minute. Interval is thepolling interval between the start time and the endtime. This field sets how often data will be polledfor storage during the collection period. Figure 7highlights the message type describe above.



Figure 6. The MATP Protocol

Figure 7. Message type in MATP



Result View-display

The Internet visualization system, is capable ofprocessing four types of user requests, which arenamely, real-time analysis, collection, elementaryanalysis and detailed analysis requests. In orderto handle these four requests, the client and theserver have each four request processing modules.

Real-time analysis concerns the current state ofan IP-based network equipment and deals withthe representation of, in the form of real-timeline graphs, the results of such analysis itemsas utilization and availability. Furthermore, if theuser wishes to obtain a document version of thegraph generated, the report function can be calledto generate a report-style result. The result view-displays of real-time analysis are shown in Figure 8.

Figure 8 is a result view-display of real-timeanalysis showing the current interface packettransmission rate on the router whose IP addressis 134.75.62.2. The top curve represents the currentratio of incoming packets into the interface, andthe bottom curve represents the current ratio ofpackets exiting the interface. The X-axis representsthe monitored time of polling in hours, minutesand seconds. The Y-axis represents the traffic ratio.

Collection demand is a function that requestscollection of data over a specified period of timefor the purpose of carrying out an accumulationanalysis of Internet equipment. Figure 9 illustratesthis function. The user interface of the collection

demand system provides the system administra-tor with such functions as setting the collectionperiod, and addition/deletion of managed sys-tems for analysis. The user is required to registerthe information on managed systems that are usedor managed for the first time. It is therefore neces-sary to pre-input the management identifier—therequest ID—as well as other information on themanaged systems such as IP address, name of com-munity, line speed and period of collection beforemaking a request to the server. For managementsetting already established, there is a convenientfeature of deleting specific management identi-fiers. Figure 9 is a result of a collection request onIP addresses 203.252.53.3 to 203.252.53.9.

Elementary analysis aims to analyze elementarymanagement items in the Internet. To this end, thedata accumulated through a collection request areanalyzed by item of analysis on a elementary leveland then are presented to the user in a Web pageformat. For a more detailed result, detailed analysiscan be used. Figure 10 shows the result view-display of a request for an elementary analysis.

Figure 10 is a result view-display of an ele-mentary analysis request on a router whose IPaddress is 134.75.62.2. The results are presentedto the user in the form of Web page and includethe downcount, availability and the correspondingline speed of each and every managed system con-tained in the selection request ID. Items of analysisor managed systems marked in red indicate that

Figure 8. Example of a result view-display of a real-time analysis



Figure 9. Example of an user interface display for a collection demand

Figure 10. Example of a result display for an elementary analysis



Figure 11. Example of a result view-display for a detailed analysis

precautions are needed on the part of the user, andsuggests a need to run a detailed analysis of theseitems or systems.

In detailed analysis, the data accumulatedthrough a collection request are analyzed by itemof analysis on a comprehensive level and then arepresented to the user in the form of a graph ortable for an analysis period desired by the user. Aresult view-display of a detailed analysis is shownin Figure 11.

Figure 11 is an output of the results of incomingpacket traffic analysis on the above-managedsystem. The results are an arithmetic mean ofanalysis over a duration specified by the user andshow the discard, error, unknown protocol andnormal state ratios of packets heading into theinterface.

ConclusionThis paper makes use of SNMP MIB-II in

carrying out a performance analysis of the Internet.Specifically it utilizes Internet performance para-meters to extract such items of analysis as real-time,elementary and detailed analyses.

Furthermore, in order to provide the four userservices of real-time analysis, collection demand,

elementary analysis and detailed analysis, aWeb-based interface of an Internet performancemanagement system whereby the requests forservice are processed, functions are carried outaccording to the item of analysis and the respec-tive results are shown to the user either in theform of a graph or table, was realized with theassistance of Web-related technology and JAVA.The Web interface assumes a client/server struc-ture with the visualization client being responsiblefor the functions of managing the connectionwith the visualization server, processing user mes-sages, processing of collection/analysis requests,processing of graphs, and processing of reports.By contrast, the visualization server, in orderto process the user’s request obtained throughthe visualization client, assumes the responsibilityof carrying out SNMP management, connectionmanagement, message processing, processing ofcollection/analysis results, processing of reports,processing of items of analysis, SPES manage-ment, image generation, Web page creation andmanagement information control. Moreover, toensure a reliable communication between the visu-alization client, and the visualization server andbetween the visualization server and the accumu-lation analysis items, a protocol by the name of



MATP was designed for information transmissionfunctions. The results obtained through the sys-tem are presented via a result on-screen displayand an analysis report. Furthermore, managementwith existing equipment and at no additional costwas possible with the use of MIB-II, an Internetmanagement standard.

In conclusion, the research presented in thispaper made it possible to realize a platform-independent management of the Internet by defin-ing items of analysis using SNMP MIB-II and bydesigning and implementing a Web-based userinterface of an Internet management system capa-ble of analyzing the above items of analysis.

References1. Blommers J. Practical Planning for Network Growth,

Prentice Hall, Englewood Cliffs, NJ, 1996.2. Deri L. Surfin’ Network Resources Across the Web, IBM

Zurich Resezrch Laboratory, 1996.3. Cornell G, CAY S, Horstmann CS. Core Java, Prentice

Hall, Englewood Cliffs, NJ, 1996.4. Hughes, Merlin H. JAVA Network Programming,

Manning Publications, New York.

5. Koo JH. A Study of the design of Web-based interfacefor network traffic analysis system. The GraduateSchool of Sung Kyun Kwan University, 1997.

6. Havel O, Patel A. Design and implementationof a composite performance evaluation model forheterogeneous network management applications,1995.

7. Han J-S, Ahn SJ, Chung JN. Web-based perfor-mance manager system for a Web server. Net-work Operations And Management Symposium ’98,1998.

8. Shin S-C, Ahn SJ, Chung JW. Design and Imple-mentation of SNMP-based Performance parameterextraction system. 1997 Asia-Pacific Network Oper-ations and Management Symposium, 1997.

9. Shin S-C, Ahn SJ, Chung JW. A new approach togather network management data periodically. ITC-CSCC 97, 1997.

10. Stallings W. SNMP, SNMPv2, and CMIP: The PracticalGuide to Network-Management Standards, Addison-Wesley, Reading, MA, 1993. �

If you wish to order reprints for this or anyother articles in the International Journal ofNetwork Management, please see the SpecialReprint instructions inside the front cover.


design and implementation of a web-based internet performance

Documents