Jepsen Theo, Moshref Masoud, Carzaniga Antonio, Foster Nate, Soulé Robert (2018), Packet Subscriptions for Programmable ASICs, in the 17th ACM Workshop
, Redmond, WA, USAACM, Redmond, WA.
JepsenTheo, AlvarezDaniel, FosterNate, KimChanghoon, LeeJeongkeun, Fast String Searching on PISA, in Symposium on SDN Research (SOSR)
, ACM SIGCOMM, San Jose, CA.
Jepsen Theo, Moshref Masoud, Carzaniga Antonio, Foster Nate, Soule Robert, Life in the Fast Lane: A Line-Rate Linear Road, in 2018 Symposium on SDN Research
, ACM, Symposium on SDN Research (SOSR).
Jin Xin, Li Xiaozhou, Zhang Haoyu, Soule Robert, Lee Jeongkeun, Foster Nate, Kim Changhoon, Stoica Ion, NetChain: Scale-Free Sub-RTT Coordination, in 15th USENIX Symposium on Networked Systems Design and Implementation
, USENIX, Symposium on Networked Systems Design and Implementation.
Our modern world is flooded with data. An incessant amount of data, generated by people, software systems, and the physical world is more accessible than ever before and is much larger in volume, variety, and velocity. In many application domains, live data recording the interactions between people, systems, and the environment is available for analysis. This data often takes the form of an interaction graph, a temporally evolving graph, where entities are the vertices and the interactions between them are the edges. Examples of interaction graphs included phone call-detail records which log data about phone calls, or tweets which record the postings of users over Twitter. Given the massive size of the data sets available, analysts often find themselves drowning in a sea of data. To make sense of it, users typically would like to proceed in an ad-hoc or exploratory fashion. That is, they formulate conjectures about trends in the data, test those conjectures, and then based on the result, rule out hypotheses and generate new ones. This process is roughly analogous to how programmers understand large, complex systems. Unfortunately, while programmers have tools and visual IDEs which help them understand code, they lack the appropriate tools that would allow them to gain insights about very large graph data sets.Interaction graphs have several characteristics that make their processing and analysis challenging. First, the number of edges in an interaction graph can grow to an unbounded size. Thus, analyzing such graphs requires techniques that scale to massively sized data sets. Second, interaction graphs are temporal, i.e., they evolve over time. Understanding how the data changes (again at scale) requires the conception and development of novel processing techniques. Third, the analytics performed on interaction graphs typically involve computations on data stored as edge attributes. Thus, while most graph processing systems focus on the structure of the data, with interaction graphs operations are needed that bear a stronger resemblance to traditional relational database queries. Fourth, understanding interaction graphs, given the vast amounts of evolving information, is truly one of the grand challenges of data science, and requires novel techniques that involve the modeling and interactive visualization the data.The goal of this collaborative project is to conceive, develop, and validate novel techniques and ap- proaches to enable real-time analytics of interaction graphs, by bringing together the skills and know-how of researchers with diverse backgrounds.