next up previous
Next: NetPIPE Design Up: NetPIPE Previous: NetPIPE

Introduction

In recent years, much research has been directed towards evaluating the performance of high speed networks. [2,3,4,5] The design of NetPIPE, a network protocol independent performance evaluator, has been motivated by the need to assess the performance of communication bound applications. NetPIPE helps answer questions that surround network communications inherent to these applications. These applications include file transfer and graphical simulations for display in a virtual reality environment, such as CAVE [13] applications, which require frame transfers from a compute server. While file transfer applications allow streaming of data, a graphical simulation requires blocks of data transmitted at regular intervals to maintain full-motion video. The size of each block and the number of frames per second are enough to specify a minimum network throughput required to maintain realistic animation.

With the applications in mind, several questions can be asked in reference to the network communication. For instance, how soon will a given data block of size k arrive at its destination? Which network and protocol will transmit size k blocks the fastest? What is a given network's effective maximum throughput and saturation level? Does there exist a block size k for which the throughput is maximized? How much communication overhead is due to the network communication protocol layer(s)? How quickly will a small (< 1 kbyte) control message arrive, and which network and protocol are best for this purpose?

The answers to such questions are not always straightforward and easy to obtain with current network performance tools. The two most popular tools, ttcp [3] and netperf [2], are based on the TCP/IP [7,8,9] communications protocol. While netperf has the ability to map network performance, comparing network protocols with these tools is difficult if not impossible. Finding the effective maximum bandwidth using ttcp is an exercise in delving into protocol internals. Knowledge of the appropriate buffer size, alignment address, and protocol settings is required to achieve data transfer at the effective maximum bandwidth.

With the various network types available (ATM, FDDI, HIPPI, Ethernet, etc.), it is difficult to select a network infrastructure which best satisfies an application's bandwidth requirement. The design of NetPIPE has been motivated by the need to select a network infrastructure for various types of applications and communication with a CAVE virtual reality environment. In addition NetPIPE provides for visualization of network performance and the information necessary to answer the above questions.

This paper presents NetPIPE and some of the results obtained through its use. In the next section, we present the NetPIPE driver and its underlying principles. Sections 3 and 4 consist of results obtained using NetPIPE in a variety of network infrastructures. A summary and conclusion with answers to the questions posed above can be found in Section 5.



next up previous
Next: NetPIPE Design Up: NetPIPE Previous: NetPIPE



Quinn Snell
Fri Apr 5 12:14:32 CST 1996