Projects

Prepositioning Content to Clients for Faster Page Loads

Prepositioning content on browsers before they request it can reduce page load times. While prepositioning mechanisms exist in modern browsers, effective policies for prepositioning are less well understood. In this work, we explore policies for prepositioning content for websites hosted by large CDNs. In these CDNs, prepositioning has the potential to be effective for two reasons: the HTML is often dynamically generated at an origin server, introducing a window of time during which a CDN proxy can preposition content; and, CDNs have instrumentation using which we can accurately estimate the set of objects to preposition. We devise policies for prepositioning objects before the arrival of the HTML at the proxy, and before the user navigates to the next page within the site. Our results, using live downloads from 160 sites hosted on a large global CDN, show that, with careful prepositioning, we can achieve, in the median, 50% of the performance benefit of having the content cached on the browser.

Modeling HTTP/2 Speed from HTTP1.1 Traces

With the impending standardization of HTTP/2, content providers would like to understand the benefits and pitfalls of transitioning to the new standard. Prior research has reported mixed results with SPDY, the precursor to HTTP/2. Using a large dataset of HTTP/1.1 resource timing data from production traffic on large CDN, and a model of HTTP/2 behavior, we obtain the distribution of performance differences between the protocol versions for nearly 280,000 downloads. We find that HTTP/2 provides significant performance improvements in the tail, and, for websites for which HTTP/2 does not improve median performance, we explore how, and for which websites, optimizations like prioritization and push can improve performance, and how these improvements relate to page structure. Finally, we explore the efficacy of previously proposed TCP enhancements to ameliorate the adverse effects of loss on HTTP/2 performance.

Diagnosing Mobile Path Inflation

As mobile Internet becomes more popular, carriers and content providers must engineer their topologies, routing configurations, and server deployments to maintain good performance for users of mobile devices. Understanding the impact of Internet topology and routing on mobile users requires broad, longitudinal network measurements conducted from mobile devices. In this work, we are the first to use such a view to quantify and understand the causes of geographically circuitous routes from mobile clients using 1.5 years of measurements from devices on 4 US carriers. We identify the key elements that can affect the Internet routes taken by traffic from mobile users (client location, server locations, carrier topology, carrier/content-provider peering). We then develop a methodology to diagnose the specific cause for inflated routes. Although we observe that the evolution of some carrier networks improves performance in some regions, we also observe many clients – even in major metropolitan areas – that continue to take geographically circuitous routes to content providers, due to limitations in the current topologies.

Towards Loss-less Data Center Networks

Data centers must support a range of workloads with differing demands. Although existing approaches handle routine traffic smoothly, intense hotspots – even if ephemeral– cause excessive packet loss and severely degrade performance. This loss occurs even though congestion is typically highly localized, with spare buffer capacity at nearby switches. In this paper, we argue that switches should share buffer capacity to effectively handle this spot congestion without the monetary hit of deploying large buffers at individual switches. Specifically, we present detour-induced buffer sharing (DIBS), a mechanism that achieves a near lossless network without requiring additional buffers at individual switches. Using DIBS, a congested switch detours packets randomly to neighboring switches to avoid dropping the packets. We implement DIBS in hardware, on software routers in a testbed, and in simulation, and we demonstrate that it reduces the 99th percentile of delay-sensitive query completion time by up to 85%, with very little impact on other traffic.

Publications

Kyriakos Zarifis, Mark Holland, Manish Jain, Ethan Katz-Bassett, Ramesh Govindan, Making Effective Use of HTTP/2 Server Push in Content Delivery Networks, Technical Report, University of Southern California, May 2017 [PDF] 

Kyriakos Zarifis, Mark Holland, Manish Jain, Ethan Katz-Bassett, Ramesh Govindan, Modeling HTTP/2 Speed from HTTP/1 Traces, Passive and Active Measurement Conference (PAM ’16), March 2016. [PDF]

Brandon Schlinker, Kyriakos Zarifis, Italo Cunha, Nick Feamster, Ethan Katz-Bassett, PEERING: An AS for Us, Proceedings of the 13th ACM Workshop on Hot Topics in Networks (HotNets), October 2014. [PDF]

Brandon Schlinker, Kyriakos Zarifis, Italo Cunha, Nick Feamster, Ethan Katz-Bassett, Minlan Yu, Try Before you Buy: SDN Emulation with (Real) Interdomain Routing, Proceedings of the 2014 Open Networking Summit (ONS), published by USENIX, March 2014. [PDF]

Kyriakos Zarifis, Rui Miao, Matt Calder, Ethan Katz-Bassett, Minlan Yu, Jitendra Padhye, DIBS: Just-in-time Congestion Mitigation for Data Centers, Ninth Eurosys Conference 2014, EuroSys 2014, April 2014. [PDF]

Kyriakos Zarifis, Tobias Flach, Srikanth Nori, David Choffnes, Ramesh Govindan, Ethan Katz-Bassett, Z. Morley Mao, Matt Welsh, Diagnosing Path Inflation of Mobile Client Traffic, Passive and Active Measurement Conference (PAM ’14), March 2014. [PDF]

Brandon Schlinker, Kyriakos Zarifis, Italo Cunha, Nick Feamster, Ethan Katz-Bassett, Minlan Yu, Towards Impactful Routing Research: Running Your Own (Emulated) AS on the (Real) Internet, Proceedings of the ACM CoNEXT Student Workshop, December 2013. [PDF]

Kyriakos Zarifis, Tobias Flach, Srikanth Nori, David Choffnes, Ramesh Govindan, Ethan Katz-Bassett, Morley Mao, Matt Welsh, Diagnosing Path Inflation of Mobile Clients, No. 13-934, June 2013. [PDF]