Browsing by Author "Maggs, Bruce Macdowell"

Delivering content of interest to clients is one of the most important tasks of the Internet

and an everlasting research question of today's networking. Content distribution networks(CDNs)

emerged in response to the rising demand of content providers to deliver contents to clients efficiently,

reliably, and securely at relatively low cost.

This dissertation explores how CDNs can achieve major performance benefits by adopting better

caching strategies without changing the network, or by collaboration with ISPs and taking advantage of their

better knowledge of network status and topology. It discusses the emerging trends of hybrid CDN architectures

and solutions to reliability problems introduced by them. Finally, it demonstrates how CDNs could better

protect both content providers and consumers from attacks and other malicious behaviors.

Providing a satisfactory quality of experience (QoE) to Internet users is crucial for content and service providers. When users get bad QoE from an application, such as the videos they are watching on a streaming provider keep freezing or the shopping Web site they are visiting takes a long time to load,

they often spend less time on the application, return to it less frequently,

or even worse they might switch to an alternative application, in all cases hurting the business financially.

While some applications such as Internet telephony have stringent latency requirements and others such as video-on-demand are more delay tolerant and depend more on bandwidth, there are also many applications such as live streaming and Web browsing which require both low latency and sufficient bandwidth for good QoE. Towards the goal of improving application QoE in general, this thesis examines the potential of making better use of the existing Internet infrastructure in two separate directions. In one direction, we focus on reducing latency on the Internet by targeting the infrastructural inefficiencies in the lower layers of the network stack, and in the other we focus on efficient allocation of bandwidth to applications based on user preferences. With these goals in mind, this thesis attempts to answer the following questions specifically: (a) How slow is the Internet today and what are the causes of latency?

(b) Can we make better use of existing fiber-optic infrastructure to reduce latencies in the wide-area? and (c) Can we provide the users with simple mechanisms to express their application preferences and enforce those preferences in the network?

We are witnessing a paradigm shift in computing---people are increasingly using Web-based software for tasks that only a few years ago were carried out using software running locally on their computers. The increasing use of mobile devices, which typically have limited processing power, is catalyzing the idea of offloading computations to the cloud. It is within this context of cloud computing that this thesis attempts to address a few key questions: (a) With more computations moving to the cloud, what is the state of the Internet's core? In particular, do routing changes and consistent congestion in the Internet's core affect end users' experiences? (b) With software-defined networking (SDN) principles increasingly being used to manage cloud infrastructures, are the software solutions robust (i.e., resilient to bugs)? With service outage costs being prohibitively expensive, how can we support network operators in experimenting with novel ideas without crashing their SDN ecosystems? (c) How can we build a large-scale passive IP geolocation system to geolocate the entire IP address space at once so that cloud-based software can utilize the geolocation database in enhancing the end-user experience? (d) Why is the Internet so slow? Since a low-latency network allows more offloading of computations to the cloud, how can we reduce the latency in the Internet?

Accurate IP geo-location is crucial to the effectiveness of a wide array of Internet-based services ranging from targeted advertising and website localization to content delivery, security logging and authentication. The most widely used technique for remote IP geo-location is to passively query a pre-built database mapping IP blocks to physical locations. Recent analysis of commercially available databases has revealed limited global coverage and limited accuracy below the country level.

In this work, we first present a new form of geo-location technique which cross-references Regional Internet Registry (RIR) entries with topology information derived from Border Gateway Protocol (BGP) routing data. Second, we present a Hadoop integrated PATRICIA tree designed to store this dataset. Finally, we present a system for accurately and efficiently mapping location strings to representative alpha-shape polygons.

Our experiments show that cross-referencing RIR entries with topology information allows for improvements in location accuracy below the country level in comparison with traditional databases. Furthermore, we show that use of a PATRICIA tree provides maximum storage efficiency with minimal performance impact. Finally, we show that representing locations as alpha shapes provides a high level of accuracy with minimal performance overhead.

IP Anycast, as a vital routing technique, can distribute user requests to different servers with the same IP worldwide. It can improve large-scale distributed systems performance and load balance. Nonetheless, all the sites in the anycast-based system have identical IP addresses, which makes it challenging to control the system’s catchment (which site the user should go to) and results in anycast performance inefficiency.

In this thesis, we introduce two approaches to optimize the performance of IP anycast, proactively and passively. The first approach-AnyOpt, managed to build a prediction model to predict the catchment site of the user with controlled experiments and measurements with the sites. Using AnyOpt, a network operator can find a subset of anycast sites that minimizes client latency. In an experiment using 15 sites, each peering with one of six transit providers, AnyOpt predicted site catchments of 15 300 clients with 94.7% accuracy and client RTTs with a mean error of 4.6%. AnyOpt identified a subset of 12 sites, announcing to which lowers the mean RTT to clients by 33 ms compared to a greedy approach that enables the same number of sites with the lowest average unicast latency.

The second approach-regional anycast, is an approach that we found to have already been implemented by two large CDNs (Edgio and Imperva). In regional anycast, a CDN divides its content-hosting sites into different geographic regions, announces a distinct IP anycast prefix from each region, and uses DNS and IP-geolocation to direct a client to a CDN site in the same geographic area. We aim to understand how a regional anycast CDN partitions its sites and maps its customers’ clients, and how a regional anycast CDN performs compared to its global anycast counterpart. We study the deployment strategies and the performance of two CDNs (Edgio and Imperva) that currently deploy regional IP anycast. We find that both Edgio and Imperva partition their sites and clients following continent or country borders. In addition, we compare the client latency distribution in Imperva’s regional anycast CDN with that in its similar-scale DNS global anycast network, after discounting the relevant deployment differences between the two networks. We find that regional anycast can effectively mitigate the pathology in global IP anycast where BGP routes a client’s traffic to a distant CDN site (e.g., a site in a different continent). However, DNS mapping inefficiencies, where DNS returns a sub-optimal regional IP anycast address that does not cover a client’s low-latency CDN sites, can harm regional anycast’s performance. Finally, using the Tangled testbed, we show what performance benefit regional IP anycast can achieve if we discount DNS mapping sub-optimality.

We also include a measurement work about the ever-increasing anycast flipping. We observe an increase in flipping over the past several years, reaching 4.4% of RIPE Atlas vantage points in 2023. We present evidence that the prevalence of anycast flipping is increasing, and for a small but not negligible portion of clients, the impact on web performance is significant.

Novel applications have driven innovation in the Internet over decades. Electronic mail and file sharing drove research for communication and congestion control protocols. Hypertext documents then created the web and put the web browser at the center. Online advertisement commercialized the web and accelerated development in web technologies such as JavaScript along with content delivery and caching. Video streaming then demanded higher bandwidth both in the data center and the home network. The web is now headed towards increased interactivity and immersion. With high bandwidth available to many subscribers, end-to-end network latency is likely to be the bottleneck for interactive applications in the future. While some applications have very stringent latency requirements, many have a "good enough" latency floor, beyond which further speed-up is imperceptible to humans. In the latter case, time saved from reduced network latency can be used to improve other aspects of user experience. For example, most private information retrieval protocols require more computation or multiple roundtrips, and reduced network latency can allow clients to use such protocols to protect user privacy while also delivering good quality of experience. The latency budget is then set by the "good enough" latency floor (which may vary over applications). We can save by reducing network latency, and then spend to improve various aspects of the web ecosystem. This thesis (a) addresses a widespread pitfall in measuring latency on the web, and highlights that (b) there is ample potential to reduce infrastructural, long-distance latency, and (c) the saved latency enables improvements in the web ranging from increased publisher revenues for online ads to improved user privacy for DNS queries.