David Choffnes |
The Internet Measurement Conference provided a formal works-in-progess (WiP) session in November, 2016. This report describes the process for organizing and selecting participants, the format of the event, and summaries of the work presented.
In the past five years, the graduate networking course at Stanford has assigned over 200 students the task of reproducing results from over 40 networking papers. We began the project as a means of teaching both engineering rigor and critical thinking, qualities that are necessary for careers in networking research and industry. We have observed that reproducing research can simultaneously be a tool for education and a means for students to contribute to the networking community. Through this editorial we describe our project in reproducing network research and show through anecdotal evidence that this project is important for both the classroom and the networking community at large, and we hope to encourage other institutions to host similar class projects.
Download the full article DOI: 10.1145/3089262.3089266
A large body of economic research has shown the strong correlation between broadband connectivity and economic productivity. These findings motivate government agencies such as the FCC in the US to provide incentives to services providers to deploy broadband infrastructure in unserved or underserved areas. In this paper, we describe a framework for identifying target areas for network infrastructure deployment. Our approach considers (i) infrastructure availability, (ii) user demographics, and (iii) deployment costs. We use multi-objective optimization to identify geographic areas that have the highest concentrations of un/underserved users and that can be upgraded at the lowest cost. To demonstrate the efficacy of our framework, we consider physical infrastructure and demographic data from the US and two different deployment cost models. Our results identify a list of counties that would be attractive targets for broadband deployment from both cost and impact perspectives. We conclude with discussion on the implications and broader applications of our framework.
Download the full article DOI: 10.1145/3089262.3089265
Measurement has become fundamental to the operation of networks and at-scale services—whether for management, security, diagnostics, optimization, or simply enhancing our collective understanding of the Internet as a complex system. Further, measurements are useful across points of view—from end hosts to enterprise networks and data centers to the wide area Internet. We observe that many measurements are decoupled from the protocols and applications they are designed to illuminate. Worse, current measurement practice often involves the exploitation of side-effects and unintended features of the network; or, in other words, the artful piling of hacks atop one another. This state of affairs is a direct result of the relative paucity of diagnostic and measurement capabilities built into today’s network stack.
Given our modern dependence on ubiquitous measurement, we propose measurability as an explicit low-level goal of current protocol design, and argue that measurements should be available to all network protocols throughout the stack. We seek to generalize the idea of measurement within protocols, e.g., the way in which TCP relies on measurement to drive its end-to-end behavior. Rhetorically, we pose the question: what if the stack had been built with measurability and diagnostic support in mind? We start from a set of principles for explicit measurability, and define primitives that, were they supported by the stack, would not only provide a solid foundation for protocol design going forward, but also reduce the cost and increase the accuracy of measuring the network.
Download the full article DOI: 10.1145/3089262.3089264
This is a report on the Workshop on Tracking Quality of Experience in the Internet, held at Princeton, October 21–22, 2015, jointly sponsored by the National Science Foundation and the Federal Communication Commission. The term Quality of Experience (QoE) describes a user’s subjective assessment of their experience when using a particular application. In the past, network engineers have typically focused on Quality of Service (QoS): performance metrics such as throughput, delay and jitter, packet loss, and the like. Yet, performance as measured by QoS parameters only matters if it affects the experience of users, as they attempt to use a particular application. Ultimately, the user’s experience is determined by QoE impairments (e.g., rebuffering). Although QoE and QoS are related—for example, a video rebuffering event may be caused by high packet-loss rate—QoE metrics ultimately affect a user’s experience.
Identifying the causes of QoE impairments is complex, since the impairments may arise in one or another region of the network, in the home network, on the user’s device, in servers that are part of the application, or in supporting services such as the DNS. Additionally, metrics for QoE continue to evolve, as do the methods for relating QoE impairments to underlying causes that could be measurable using standard network measurement techniques. Finally, as the capabilities of the underlying network infrastructure continues to evolve, researchers should also consider how to design infrastructure and tools can best support measurements that can better identify the locations and causes of QoE impairments.
The workshop’s aim was to understand the current state of QoE research and to contemplate a community agenda to integrate ongoing threads of QoE research into a collaboration. This summary report describes the topics discussed and summarize the key points of the discussion. Materials related to the workshop are available at http://aqualab.cs.northwestern.edu/NSFWorkshop-InternetQoE
Download the full article DOI: 10.1145/3041027.3041035
In recent years, Information-centric Networking (ICN) has received much attention from both academic and industry participants. ICN offers data-centric inter-networking that is radically different from today’s host-based IP networks. Security and privacy features on today’s Internet were originally not present and have been incrementally retrofitted over the last 35 years. As such, these issues have become increasingly important as ICN technology gradually matures towards real-world deployment. Thus, while ICN-based architectures (e.g., NDN, CCNx, etc.) are still evolving, it is both timely and important to explore ICN security and privacy issues as well as devise and assess possible mitigation techniques.
This report documents the highlights and outcomes of the Dagstuhl Seminar 16251 on “Information-centric Networking and Security.” The goal of which was to bring together researchers to discuss and address security and privacy issues particular to ICN-based architectures. Upon finishing the three-day workshop, the outlook of ICN is still unclear. Many unsolved and ill-addressed problems remain, such as namespace and identity management, object security and forward secrecy, and privacy. Regardless of the fate of ICN, one thing is certain: much more research and practical experience with these systems is needed to make progress towards solving these arduous problems.
Download the full article DOI: 10.1145/3041027.3041034
A secure routing protocol represents a foundational building block of a dependable communication system. Unfortunately, currently no taxonomy exists to assist in the design and analysis of secure routing protocols. Based on the Dagstuhl Seminar 15102, this paper initiates the study of more structured approaches to describe secure routing protocols and the corresponding attacker models, in an effort to better understand existing secure routing protocols, and to provide a framework for designing new protocols. We decompose the routing system into its key components based on a functional model of routing. This allows us to classify possible attacks on secure routing protocols. Using our taxonomy, we observe that the most eective attacks target the information in the control plane. Accordingly, unlike classic attackers whose capabilities are often described in terms of computation complexity we propose to classify the power of an attacker with respect to the reach, that is, the extent to which the attacker can influence the routing information indirectly, beyond the locations under its direct control.
DNS cache plays a critical role in domain name resolution, providing (1) high scalability at Root and Top-level-domain (TLD) name servers with reduced workloads and (2) low response latency to clients when the resource records of the queried domains are cached. However, the pervasive misuses of domain names, e.g., the domains of “one-time-use” pattern, have negative impact on the effectiveness of DNS caching as the cache has been filled with those entries that are highly unlikely to be retrieved. In this paper, we investigate such misuse and identify domain name-based features to characterize those one-time domains. By leveraging the features that are explicitly available from the domain name itself, we build a classifier to combine these features, propose simple policy modifications on caching resolvers for improving DNS cache performance, and validate their efficacy using real traces.
In this paper we investigate the vulnerability of the Internet Group Management Protocol (IGMP) to be leveraged for denial-of-service (DoS) attacks. IGMP is a connectionless protocol and therefore susceptible to attackers spoofing a third-party victim’s source address in an effort to coax responders to send their replies to the victim. We find 305K IGMP responders that will indeed answer queries from arbitrary Internet hosts. Further, the responses are often larger than the requests, hence amplifying the attacker’s own expenditure of bandwidth. We conclude that attackers can coordinate IGMP responders to mount sizeable DoS attacks.
Software-defined networking (SDN) is a well-known example of a research idea that has been reduced to practice in numerous settings. Network virtualization has been successfully developed commercially using SDN techniques. This paper describes our experience in developing production-ready, multi-vendor implementations of a complex network virtualization system. Having struggled with a traditional network protocol approach (based on OpenFlow) to achieving interoperability among vendors, we adopted a new approach. We focused first on defining the control information content and then used a generic database protocol to synchronize state between the elements. Within less than nine months of starting the design, we had achieved basic interoperability between our network virtualization controller and the hardware switches of six vendors. This was a qualitative improvement on our decidedly mixed experience using OpenFlow. We found a number of benefits to the database approach, such as speed of implementation, greater hardware diversity, the ability to abstract away implementation details of the hardware, clarified state consistency model, and extensibility of the overall system.