What Lies Beneath: Understanding Internet Congestion
نویسندگان
چکیده
Developing measurement tools that can concurrently monitor congested Internet links at a large scale would significantly help us understand how the Internet operates. While congestion at the Internet edge typically arises due to bottlenecks existent at a connection’s last mile, congestion in the core could be more complex. This is because it may depend upon internal network policies and hence can reveal systematic problems such as routing pathologies, poorly-engineered network policies, or non-cooperative inter-AS relationships. Therefore, enabling the tools to provide deeper insights about congestion in the core is certainly beneficial. In this paper, we present the design and implementation of a large-scale triggered monitoring system that focuses on monitoring a subset of Internet core links that exhibit relatively strong and persistent congestion, i.e., hot spots. The system exploits triggered mechanisms to address its scalability; moreover, it automates selection of good vantage points to handle the common measurement experience that the much more congested Internet edges could often overshadow the observation for congestion in the core. Using the system, we characterize the properties of concurrently monitored hot spots. Contrary to common belief, we find that strong time-invariant hot spots reside in the Internet core — both within and between large backbone networks. Moreover, we find that congestion events at these hot spots can be highly correlated and such correlated congestion events can span across up to three neighboring ASes. We provide a root-cause analysis to explain this phenomenon and discuss implications of our findings. This work is supported by a Cisco Collaborative Research grant.
منابع مشابه
Advances in Congestion Control Algorithms in Internet
The review is done focused on Transmission Control Protocol/Internet Protocol (TCP/IP) by studying almost recently proposed algorithms to avoid congestion and improve performance. Our study in this work list many open issues that persevere in the designing, functioning and managing of the Internet. It aims at preventing incoming packet drops at gateways because of local buffer overflows. We ide...
متن کاملModeling the Internet Congestion Control Using a Smith Controller with Input Shaping*
The Internet has shown a great capability of endless growing without incurring congestion collapse. The key of this success lies on its TCP/IP congestion control algorithm. In this paper we use control theoretic analysis to model the Internet flow and congestion control as a time delay system. We show that the self-clocking principle, which is known to be a key component of any stable congestio...
متن کاملPassive Performance Monitoring and Traffic Characteristics on the SLAC Internet Border
Understanding how the Internet is used by HEP is critical to optimizing the performance of the inter-lab computing environment. Typically use requirements have been defined by discussions between collaborators. However, later analysis of the actual traffic has shown this is often misunderstood and actual use is significantly different to that predicted. Passive monitoring of the real traffic pr...
متن کاملMovement disorders: what lies beneath?
Journal of Neurology Neurosurgery & Psychiatry (ISSN No: 0022-3050) is published monthly by BMJ Publishing Group and is distributed in the USA by Air Business Ltd. Periodicals postage paid at Jamaica NY 11431 POSTMASTER: send address changes to Journal of Neurology Neurosurgery & Psychiatry, Air Business Ltd, c/o Worldnet Shipping Inc., 156-15, 146th Avenue, 2nd Floor, Jamaica, NY 11434, USA. E...
متن کامل