A Survey of Benchmarking Techniques for Network Performance. (original) (raw)
Related papers
A Distributed Instrument for Performance Analysis of Real-Time Ethernet Networks
IEEE Transactions on Industrial Informatics, 2000
Ethernet technology is widely used in real-time industrial automation. Thanks to real-time Ethernet (RTE) protocols, defined in IEC61784-2 standard, new top-performance automation solutions can be created. Such systems may have communication cycle time down to tens of s and cycle jitter less than 1 s, making network testing and debugging very critical. Existing network and protocol analyzers can perform detailed local analysis, but characterization of high-performance RTE systems requires measurement of transmission delays and these instruments cannot be adequately synchronized among them to realize a distributed measurement network. This paper introduces a new low-cost distributed measurement instrument to measure timing characteristics of RTE nodes (end-to-end delays, synchronization, etc.). The proposed instrument has multiple FPGA-based probes that allow for simultaneous/synchronized logging on different place of the target RTE network. A PC-based "monitor station" stores all the data, ready for further elaboration. Architecture details are discussed, a prototype has been realized, and some experimental results are presented. For instance, synchronization accuracy between probes is below 100 ns.
Performance Measurements of Protocols to Ethernet Real-Time Applications
2006
This paper realizes performance measurements comparing implementations of data communication with UDP and RAW protocols over Ethernet Technology, identifying the most viable alternative to support the development of distributes real-time application, searching a protocol solution to nondeterminism of bus arbitration scheme in industrial network based on Ethernet technology.
Real time characteristics of Ethernet and its improvement
Proceedings of the 4th World Congress on Intelligent Control and Automation (Cat. No.02EX527), 2002
There is a growing interest of using Ethernet to support time constrained communication raised from process control, factory automation and other real-time applications. This paper first presents a comprehensive analysis of the temporal characteristics of traditional Ethernet and the improvements for them, then summarizes the interesting features offered by a switched Ethernet for supporting time constrained communication. At last, this paper points that it still lacks of general means to handle capability of switched Ethernet after investigating QoS (Quality of Service) of switched Ethernet.
International Journal of Reconfigurable and Embedded Systems (IJRES)
Real-time communication is important in control network. In real-time communication, message need to be delivered from source to destination within specification. Embedded Ethernet and Controller Area Network (CAN) protocol can be used in control network to achieve hard real-time communication. For embedded Ethernet protocol, Carrier Sense Multiple Access with Collision Detection (CSMA/CD) is the media access control (MAC) used to control data transmission between nodes in network. Backoff algorithm in CSMA/CD is used to handle packet collisions and retransmission. For CAN protocol, it is communication protocol developed mainly for automotive application. It has priority arbitration to handle collisions and retransmission. In this project, embedded Ethernet network models and CAN network models are developed and simulated in MATLAB Simulink software. Several back-off algorithms, which are Binary Exponential Backoff (BEB), Linear Back-off Algorithm, Exponential-Linear back-off Algorithm and Logarithm Back-off Algorithm are proposed and implemented into Embedded Ethernet network model to evaluate the performance. Both embedded Ethernet and CAN network models are extended to 3 nodes, 10 nodes, and 15 nodes to evaluate performance at different network condition. The performance criteria evaluated and discussed are average delay and jitter of packets. The results show that in network with high number of nodes, Linear Back-off Algorithm and Exponential-Linear back-off Algorithm shows improvement in packets delay and jitter. For CAN network, the packet jitter is relatively low.
Performance Characteristics of Two Ethernets: An Experimental Study
Proceedings of the 1985 ACM SIGMETRICS …, 1985
Local computer networks are increasing in popularity for the interconnection of computers for a variety of applications. One such network that has been implemented on a large scale is the Ethernet. This paper describes an experimental performance evaluation of a 3 and a 10 Mb/s Ethernet. The effects of varying packet length and transmission speed on throughput, mean delay and delay distribution are quantified. The protocols are seen to be fair and stable. These measurements span the range from the region of high performance of the CSMA/CD protocol to the upper limits of its utility where performance is degraded. The measurements are compared to the predictions of existing analybical models. The correlation is found to range from good to poor, with more sophisticated models yielding better results tban a simple one.
Temporal behavior of Ethernet communications: Impact of the operating system and protocol stack
2010
Ethernet is currently the most widely used networking technology, spanning across many application domains including embedded systems. In this particular case, Ethernet is even used in many time-critical applications in which the delay induced by communication must be short and bounded. It is thus very important to understand the entire transmission process and assess its temporal behavior. There are a number of aspects to consider, including the network protocol, network topology, network elements and end devices. This paper aims at assessing the impact of the operating system and its protocol stack implementation in the end devices on the network temporal behavior. We studied four operating systems, namely a standard Ubuntu distribution with and without a real-time kernel patch, an embedded stripped down version of Linux and QNX Neutrino, and two hardware platforms, namely ordinary PCs and a single board computer based on an AVR32 CPU. We measured the Round Trip Delay (RTD) using RAW, UDP and TCP sockets to interface the protocol stack. We verified that on high computing power platforms the difference between the sockets is small but still significant in resource-constrained platforms. On the other hand, full featured general OSs present rather large worst-case delays. These can be reduced using real-time patches for those OSs, RTOSs, or even removing unnecessary modules, services and particularly, data intensive device drivers. We believe this study can be helpful for system designers as well as for teaching networks courses in embedded systems.
TCP/IP measured performance over an ethernet local network
European Transactions on Telecommunications, 1992
Both a reliable transfer and a datagram transmission protocol are provided at the transport level of the Internet protocol suite. In this paper, an evaluation of the performance of these different protocols in the presence of background load is presented. The experimental analysis was carried out on an Ethernet Local Area Network with PC-type workstations. The loading conditions were established by means of an artificial traffic generator. A reproducible experimental environment and a methodology to achieve accuracy in measurements were developed. '01. 3. No. 1
Packet Delay Analysis and Optimization for Time Management in Network Devices - Dissertation Summary
This dissertation presents the packet delay analysis for improved network delay measurement to achieve precise time synchronization in enterprise-wide network that consists of several routers, switches and computer workstations. The consideration of the probe packet design is to minimize packet delay variation on the protocol stack and be robust against cross traffic effect. In this study, we used client and server computers running on Linux version 2.6 kernel which supports the New API (NAPI) to adapt high-speed networking processes. A high-resolution low-latency packet capturing device was used to measure packet delay precisely. The experiments include packet delay measurement on the protocol stack and network devices, including several network configurations and an actual enterprise-wide network . Based on experiment results and analysis, the last packet of the packet train can be used to improve the accuracy and stability of delay measurement because they are not affected by interrupt latency and cross traffic effects. The proposed packet train probe performance was further evaluated in a networked environment to synchronize remotely located clocks. By considering the last packet, the short-term stability (based on MADEV) in 5-hop network is improved by a factor of ten than using the first packet of the packet train or a single packet.
Benchmarks for LAN performance evaluation
Communications of The ACM, 1988
A technique for quickly benchmarking the performance of local area networks (LANs) is presented. Programs which model both intermittent and constant network activity are given. BENCHMARKS FOR LAN PERFORMANCE EVALUATIOAI LARRY PRESS EVALUATION ALTERNATIVES Analytic modeling, simulation, and benchmarking are three alternative performance evaluation methods. Analytic models have been developed for the lowlevel physical and data-link layers of local area networks (LANs). They predict relative efficiency of different network topologies (bus, star, and ring), transmission media (twisted pair, broadband and baseband coaxial cable, and optical fiber), and accesscontrol protocols (Carrier Sense Multiple Access and token passing). Alternatives and analytic model results at this level are reviewed in [7]. Analytic models of complete network systems that include the characteristics of actual hardware and software and implement low-level protocols, file and other services as well as applications have not been developed. This is due to the level of complexity and the variety of configurations. Simulation might be valuable for modeling a network file system and its file-server hardware and software in addition to low-level protocols. A simulation language like Simscript II.5 [6] can be used to model entities like disk directory structures, seek-sequencing algorithms, This work was done for Apple Computer on a subcontract with the Seybold GPXIP. 01988 .4CM 0001.0782/SS/O800-1014 51.50 While we used a maximum of ten background workstations, the relative rankings of two configurations were always the same with two or more background workstations. This may not remain true for much larger networks or for different configurations. Our experience indicates, however, that at least for preliminary testing, two or three background workstations are sufficient.