Analytical and empirical studies have shown that self-similar traffic can have detrimental impact on network performance including amplified queuing delay and packet-loss ratio. On the flip side, the ubiquity of scale-invariant burstiness observed across diverse networking contents can be exploited to better design resource control algorithms. This paper explores the issue of exploiting the self-similar characteristics of network traffic in TCP congestion control.
Related white papers
Scheduling Support for Hard Real-Time Ethernet Networking
Previous results on traffic shaping on Switched Ethernet technology have demonstrated its practicability and effectiveness for hard real-time communication. With nodes dedicated to communication the application-to-application delays on a 5-node...
A Network on Chip Based Gigabit Ethernet Router Implemented on an FPGA
This paper shares the experiences gained from implementing an FPGA-based gigabit Ethernet router based on the SoCBUS network on chip architecture. The main reason for this project has been to...
Multiservice Ethernet Access
Ethernet is evolving from its origins in the local area network into a cost-effective and reliable wide area network technology. Market research shows that business customers are moving away from...
10 GbE Rising Factors Driving the Emergence of 10 Gigabit Ethernet
IT capacity and demand are on a collision course for the datacenter. Industry trends such as Chip Multithreading (CMT), multicore CPUs such as CoolThreads, Intel Xeon, and AMD Opteron, blade...
Accelerating Secure Business Applications Podcast
Download this Riverbed Connect podcast and listen as Bob Gilbert discusses with Mark Day, Riverbed's Chief Scientist, the details involving SSL encryption, SSL's impact on WAN optimization, and new wide-area...
Riverbed Raises the Ante Again in WDS with RiOS 5.0
Do your remote workers have a desire for higher performance computing? In this Feature Brief, learn how Steelhead appliances from Riverbed, with the enhanced RiOS 5.0 operating system, provide a...
Real-Time Collaborative Modelling Over Low-Bandwidth Networks
ModellingSpace is an open learning environment, which permits building of models by collaborating partners in various educational settings. This paper describes briefly the main features of the ModellingSpace environment and...

