<html><body><div id="zimbraEditorContainer" style="font-family: garamond,new york,times,serif; font-size: 12pt; color: #000000" class="17"><div></div><div data-marker="__QUOTED_TEXT__"><p></p><p style="margin-left: 50px;" data-mce-style="margin-left: 50px;">Arpit Gupta will present his FPO "Flexible and Scalable Systems for Network Management" on Friday, 8/31/2018 at 10 am in CS 302.<br></p><p style="margin-left: 50px;" data-mce-style="margin-left: 50px;"><span style="font-size: 12pt;" data-mce-style="font-size: 12pt;">The members of his committee are as follows: Adviser: Nick Feamster; Readers: Nick Feamster, Jennifer Rexford, and Walter Willinger (NIKSUN Inc.); Examiners: Marshini Shetty, Kyle Jamieson, and Nick Feamster.</span></p><p style="margin-left: 50px;" data-mce-style="margin-left: 50px;"><span style="font-size: 12pt;" data-mce-style="font-size: 12pt;">Everyone is invited to attend. A copy of his thesis is available in CS 310.</span></p><p style="margin-left: 50px;" data-mce-style="margin-left: 50px;"><span style="font-size: 12pt;" data-mce-style="font-size: 12pt;">Thesis abstract follows below.</span></p><p></p><div data-marker="__QUOTED_TEXT__">Our daily lives are heavily reliant upon Internet-connected devices, services, and applications.</div><div data-marker="__QUOTED_TEXT__">This reliance makes it more critical than ever that the underlying networks</div><div data-marker="__QUOTED_TEXT__">they depend on be reliable, performant, and secure. At the same time, the increasing</div><div data-marker="__QUOTED_TEXT__">complexity and diversity of today's devices, services, and applications have made</div><div data-marker="__QUOTED_TEXT__">network management tasks more complicated than ever. Modern network management</div><div data-marker="__QUOTED_TEXT__">mandates that operators can systematically monitor what is going on in their</div><div data-marker="__QUOTED_TEXT__">networks (network monitoring) and use this information to take real-time preventive</div><div data-marker="__QUOTED_TEXT__">or corrective actions (network control). Achieving these goals while also adhering to</div><div data-marker="__QUOTED_TEXT__">the limited compute and storage resources available on modern network devices poses</div><div data-marker="__QUOTED_TEXT__">signi cant challenges.</div><div data-marker="__QUOTED_TEXT__">The contribution of this dissertation is the design and implementation of two systems</div><div data-marker="__QUOTED_TEXT__">that enable exible and scalable network monitoring and control. The networkmonitoring</div><div data-marker="__QUOTED_TEXT__">system, Sonata, collects and analyzes network trac to infer various network</div><div data-marker="__QUOTED_TEXT__">events in real time. The network-control system, SDX, enables ne-grained</div><div data-marker="__QUOTED_TEXT__">reactive control actions for interdomain trac without disrupting the existing routing</div><div data-marker="__QUOTED_TEXT__">protocols. For each of these two systems, the dissertation focuses on (i) the ab-</div><div data-marker="__QUOTED_TEXT__">stractions that allow network operators to express </div><div data-marker="__QUOTED_TEXT__">exible programs for both network</div><div data-marker="__QUOTED_TEXT__">monitoring and control; (ii) the algorithms that make the best use of limited compute</div><div data-marker="__QUOTED_TEXT__">and storage resources; and (iii) the systems that combine the high-level abstractions</div><div data-marker="__QUOTED_TEXT__">and the low-level algorithms and can be deployed in production settings.</div><div data-marker="__QUOTED_TEXT__">The lessons learned from this dissertation can help us design next-generation<br></div><div data-marker="__QUOTED_TEXT__">network-management systems. More concretely, unlike existing systems that rely</div><div data-marker="__QUOTED_TEXT__">solely on a single device-type, this dissertation shows that designing systems that can</div><div data-marker="__QUOTED_TEXT__">pool resources from a heterogeneous set of devices (targets) is critical for building</div><div data-marker="__QUOTED_TEXT__">exible and scalable network-management systems. It also demonstrates that as the</div><div data-marker="__QUOTED_TEXT__">networking technologies and protocols evolve rapidly with time, it is imperative to</div><div data-marker="__QUOTED_TEXT__">design modular systems that can swiftly catch up with these changes. Finally, this</div><div data-marker="__QUOTED_TEXT__">research also illustrates that it is crucial to select strategic locations (e.g., Internet exchange</div><div data-marker="__QUOTED_TEXT__">points) for deployment to drive innovations in Internet-wide trac monitoring</div><div data-marker="__QUOTED_TEXT__">and control.</div><div data-marker="__QUOTED_TEXT__"><br></div></div></div></body></html>