Mario A. Sánchez, John S. Otto, and Zachary S. Bischof, Northwestern University; David R. Choffnes, University of Washington; Fabián E. Bustamante, Northwestern University; Balachander Krishnamurthy and Walter Willinger, AT&T Labs—Research
Abstract:
We present Dasu, a measurement experimentation platform for the Internet’s edge. Dasu supports both controlled network experimentation and broadband characterization, building on public interest on the latter to gain the adoption necessary for the former. We discuss some of the challenges we faced building a platform for the Internet’s edge, describe our current design and implementation, and illustrate the unique perspective it brings to Internet measurement. Dasu has been publicly available since July 2010 and has been installed by over 90,000 users with a heterogeneous set of connections spreading across 1,802 networks and 147 countries.
USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.
BibTeX
@inproceedings {180331,
author = {Mario A. S{\'a}nchez and John S. Otto and Zachary S. Bischof and David R. Choffnes and Fabi{\'a}n E. Bustamante and Balachander Krishnamurthy and Walter Willinger},
title = {Dasu: Pushing Experiments to the {Internet{\textquoteright}s} Edge},
booktitle = {10th USENIX Symposium on Networked Systems Design and Implementation (NSDI 13)},
year = {2013},
isbn = {978-1-931971-00-3},
address = {Lombard, IL},
pages = {487--499},
url = {https://www.usenix.org/conference/nsdi13/technical-sessions/presentation/sanchez},
publisher = {USENIX Association},
month = apr
}
Dasu is a platform for measurement-based networking experiments that are run on the machines of volunteering end-users. As such, it provides an excellent vantage point for researchers to examine the rarely studied edges of the Internet. The platform is enabled by a clever use of incentives: in exchange for donating compute and networking resources to experiments, users are offered a characterization of the service they receive from their ISP (this aspect of Dasu is covered in another publication).
This paper focuses on the mechanics of the platform: how to ensure that the experiments are useful and sound in a dynamic and volatile measurement environment, how to guarantee the safety and security of end-user machines, and how to avoid adverse impact on the network of large numbers of concurrent experiments. To date Dasu has been installed by over 90,000 users across the globe, covering over 1000 networks, and the paper demonstrates the usefulness of this deployment with three case study experiments. These case studies look at routing asymmetry, AS-level peering arrangements, and the impact of a proposed DNS extension across different geographic regions. They show that Dasu's simple rule-based, declarative programming interface, in conjunction with a small set of standard measurement utilities, can enable some powerful experiments that could significantly improve our visibility into and understanding of the Internet.
In the opinion of the PC, the great strength of this paper is that it describes a practical, real system that has a large deployment and has obvious utility. The ability to conduct coordinated experiments across sets of heterogeneous end-hosts and diverse networks is potentially very exciting to Internet researchers. The downside is that currently the platform is not generally available to other researchers, although the client software is open source and available for download.
As a research contribution, this paper doesn't contain new insights or describe novel design features. However, it does provoke thought about the possibilities for observing the Internet in new ways with this expanded and more diverse measurement base. Perhaps in the not-too-distance future Dasu will be the infrastructure that enables researchers to opportunistically analyze events such as Internet outages from a previously inaccessible viewpoint.
connect with us