Matthew Mathis

Learn More
In this paper, we analyze a performance model for the TCP Congestion Avoidance algorithm. The model predicts the bandwidth of a sustained TCP connection subjected to light to moderate packet losses, such as loss caused by network congestion. It assumes that TCP avoids retransmission timeouts and always has sufficient receiver window and sender data. The(More)
With the growth of high performance networking, a single host may have simultaneous connections that vary in bandwidth by as many as six orders of magnitude. We identify requirements for an automatically-tuning TCP to achieve maximum throughput across all connections simultaneously within the resource limits of the sender. Our auto-tuning TCP implementation(More)
TCP has become the dominant protocol for all network data transport because it presents a simple uniform data delivery service which is sufficient for most applications over all types of lower network layers. By its very nature, TCP's adaption and retransmission strategies hide all of the details of the lower layers from the application. For example the(More)
Many high performance distributed applications require high network throughput but are able to achieve only a small fraction of the available bandwidth. A common cause of this problem is improperly tuned network settings. Tuning techniques, such as setting the correct TCP buffers and using parallel streams, are well known in the networking community, but(More)
Relentless congestion control is a simple modification to AIMD congestion control: instead of halving cwnd after a loss, cwnd is reduced by the number of lost segments. It can be modeled as a strict implementation of Van Jacobson’s Packet Conservation Principle. Relentless congestion control has exactly unity gain, which is expected to make it much easier(More)
The current Internet fairness paradigm mandates that all protocols have equivalent response to packet loss and other congestion signals, allowing relatively simple network devices to attain a weak form of fairness by sending uniform signals to all ows. Our paper[1], which recently received the ACM SIGCOMM Test of Time Award, modeled the reference(More)