TCP/IP protocol stack analysis using MENet

As computing systems grow in complexity, the cluster and gird communities require more sophisticated tools to diagnose, debug and analyze such systems. Current network monitoring tools capture network traffic as it appears on the network. These tools are incapable of capturing traffic as it progresses through a running protocol stack. The monitoring tools cannot lend insight to the modulating behavior of TCP (transmission control protocol) UDP (user datagram protocol) that have been shown to dominant the network traffic and impact network performance. The current generation of tools cannot record true application traffic demands and cannot capture protocol state information with respect to the TCP/IP stack at run time in order to help fine tune network performance. The developed tools focus more on presenting the statistics like IP addresses, higher level protocol, host etc. In this paper we introduce a MENet (monitor for Ethernet network traffic). The toolkit is an extension to the earlier developed utility for traffic monitoring. The toolkit is operational on the platform of Windows NT/2000. The toolkit is developed by programming in Microsoft Visual C++ environment. The MENet decrypts the packets passing through a system. We first present the design of the toolkit and then demonstrate the implementation with screen shots taken on real time traffic. Next we evaluate the working of the toolkit, its performance and its potential use in application such as protocol tuning a traffic characterization. The protocol analysis demonstrates the effectiveness of MENeT for monitoring. The work on presenting the stats to portray the network traffic from every angle is in place.