Browsing by Subject "Network"

Now showing 1 - 17 of 17

A DoS attack mitigation strategy for software defined networks using Frenetic
(2017-05) Carvajal Barboza, Erick Mauricio; Abraham, Jacob A.
Despite efforts to improve its robustness, the Internet still has important vulnerabilities that can be easily exploited. Denial-of-Service (DoS) attacks are a good example; this kind of attack has the potential to bring down large portions of the Internet by overwhelming the network with a huge amount of useless traffic. One of the possibilities to increase network robustness is a software defined networking (SDN) approach. This type of networking has a logically centralized controller that contains information about the state of the entire network, allowing it to make better decisions than when the network state is distributed between all the switches. In this type of network, is easier for the network manager to detect and mitigate an attack. This thesis presents a DoS mitigation technique that detects and stops such attacks on hosts providing services on a software defined network. Protection of the network infrastructure itself is not targeted by our approach. This mitigation technique was implemented using Frenetic, a domain specific programming language that allows easy programming of the switches using the OpenFlow protocol. It was tested using Mininet, a widely known network emulator for SDN, and several combinations of traffic and network size were used. The strategy presented in this thesis manages to deal with all the different configurations with a reaction time lower than other approaches targeting the same problem.
Automatic semiconductor wafer map defect signature detection using a neural network classifier
(2010-12) Radhamohan, Ranjan Subbaraya; Ghosh, Joydeep; El-Hamdi, Mohamed
The application of popular image processing and classification algorithms, including agglomerative clustering and neural networks, is explored for the purpose of grouping semiconductor wafer defect map patterns. Challenges such as overlapping pattern separation, wafer rotation, and false data removal are examined and solutions proposed. After grouping, wafer processing history is used to automatically determine the most likely source of the issue. Results are provided that indicate these methods hold promise for wafer analysis applications.
Bus network redesigns in medium-sized cities : an equity evaluation on supermarket accessibility
(2021-07-30) McGee, Jordan Kathleen; Karner, Alex; Wegmann, Jake
Transit agencies across the country are redesigning their bus network for the first time in several decades in order to increase ridership and the attractiveness of the system. The reallocation of resources and resulting service cuts raise equity concerns. This report calculates and evaluates equity-focused performance measurements related to supermarket accessibility before and after the bus network redesigns in the medium-sized cities of Austin, Columbus and Indianapolis. The performance measures related to grocery store accessibility significantly improved under Columbus’ bus network redesign and appeared equitable. The measures for Indianapolis largely worsened, but people of color equitably fared better than white residents. Austin’s bus network redesign had mixed performance and equity results for grocery store accessibility. On average, the redesigns of the three bus networks did not raise significant equity concerns for grocery store accessibility.
Carbon capture and storage network optimization under uncertainty
(2018-05) Tutton, Peter Mark; Leibowicz, Benjamin D.; Hovorka, Susan D. (Susan Davis)
Carbon capture and storage is a method for emissions reductions that can be applied to both the electric sector and industrial sources. Significant uncertainties surround the technologies, policy and extent to which CCS will be deployed in the future. For widespread deployment, future CCS demand should be considered during infrastructure planning. This study presents a novel model that considers spatial information and uncertainty in generating an optimal CCS network. The two-stage stochastic model, utilizes both geographic information systems (GIS) and mixed integer programming (MIP), to generate an optimal near-term hedging strategy. The strategy considers one discrete uncertainty distribution: the future demand for CO₂ storage. A case study in the Texas Gulf Coast demonstrates the value of considering uncertainty of future demand. The optimal solution is selected from a candidate network consisting of twelve sources and five reservoirs that can be linked via a network of pipelines and ship routes. The results demonstrate that optimal hedging strategies lead to transportation cost savings of up to 14% compared to a ‘naive approach’ in which only the expected value is considered. The transportation selection also highlights the benefit of utilizing ship transport in uncertain scenarios due to their ability to be reassigned to a different route or sold.
Collaboration within the sport-based youth development non-profit network in Austin, TX
(2016-05) Klaic, Darija; Dixon, Marlene A., 1970-; Sparvero, Emily Suzanne, 1975-
This qualitative study assessed collaboration within the sport-based youth development non-profit network in Austin, TX. Network, capital, resource sharing and collaboration theories were used as lenses for this research project. Qualitative methods applied were surveys and follow-up interviews. Surveys were sent to 13 identified non-profit organizations in Austin, TX that use sports programming for youth development in order to gain insight into their structure and organization, including collaboration and partnerships. Follow-up interviews were recorded, transcribed, coded and analyzed. Findings uncovered that there is no collaboration between the organizations participating in the study, but that their respective cross-sectoral collaboration networks are of vital importance to the organizations’ existence and programming. Recommendations were made on future collaborations within the network and possible benefits of forming a coalition were discussed.
Evaluation of open-source intrusion detection systems for IPv6 vulnerabilities in realistic test network
(2017-05-03) Gin, Jeremy; Evans, Brian L. (Brian Lawrence), 1965-; Bard, William C
The Internet Protocol (IP) defines the format by which packets are relayed throughout and across networks. A majority of the Internet today uses Internet Protocol version 4 (IPv4), but due to several key industries, a growing share of the Internet is adopting IPv4’s successor, Internet Protocol version 6 (IPv6) for its promise of unique addressability, automatic configuration features, built-in security, and more. Since the invention of the Internet, network security has proven a leading and worthwhile concern. The evolution of the information security field has produced an important solution for network security monitoring: the intrusion detection system (IDS). In this report, I explore the difference in detection effectiveness and resource usage of two network monitoring philosophies, signature-based and behavior-based detection. I test these philosophies, represented by leading edge passive monitors Snort and Bro, against several categories of state-of-the-art IPv6 attacks. I model an IPv6 host-to-host intrusion across the Internet in a virtual test network by including benign background traffic and mimicking adverse network conditions. My results suggest that neither IDS philosophy is superior in all categories and a hybrid of the two, leveraging each’s strengths, would best secure a network against leading IPv6 vulnerabilities.
Feeding the technopolis : an overview of the potential emergence of homemade food markets in neighborhoods and its energy savings : a trans-disciplinary approach
(2018-12) Azagra, Marcelo; King, Carey Wayne, 1974-
Individual decision making (e.g. deciding which type of food to eat) has recently significantly benefited from the widespread use of machine learning in artificial intelligence (AI) applications (e.g. search-friendly metadata) used by companies like Yelp, Google, Amazon, and Facebook. Nevertheless, certain goals are difficult to achieve via the collective use of AI. As a result, the centralized authority framework of control prevails, governing the factors of collective intelligence (e.g. countries, cities, neighborhoods, groups). AI gets its value from large data sets obtained from exploiting collective human interactions, mathematical models, and computing power. So, while AI is usually compared with human intelligence (e.g. Deep Blue beat chess' master in 1997 and AlphaGo beat Go's master in 2017), both complement each other more than is usually admitted. The emergence of new ideas often mimics the way natural systems survive and evolve. Some living organisms such a slime mold, ant colonies, and beehives work collectively especially when food is scarce. For instance, ant colonies follow rules written in the form of pheromones with different intensities, a “chemical alphabet” to signal specific requests and warnings, ultimately creating an autocatalytic processes that functions as a collective brain to find food and survive. By extension, the social and economic value of technological innovations for humans is comparable. Linking previously disconnected agents through electronic platforms increases the number of interactions and the number of choices available, minimizing their transaction cost. Moreover, the blurring roles of actors interacting in the cyberspace and physical space are a distinctive characteristic of the new economy (e.g. Amazon’s acquisition of Whole Foods). People work collectively producing, consuming, and improving solutions aided by Peer-to-Peer (P2P) networks, enabling the rise of self-management systems without central planning (e.g. Wikipedia, DIY bio). Democracy and human rights are notable examples of social innovations that historically challenged the status-quo. The rise of self-organizing political and economic structures without a central authority could be the next one. Moreover, if technologies are configured to enable regular, honest, and cooperative behavior through social norms and programmable trust, it can be a powerful tool for the emergence of new collective actions without central planning in neighborhoods, cities, and countries. One of the affordances of AI applications is to solve problems that scale globally such as energy and food security. AI applications are helpful, for example, by providing more efficient grid operation or an optimal inventory and delivery management system. They also reduce waste and energy consumption by supporting the operation of networked collaborative systems capable of changing the very nature of food consumption habits and improving the overall efficiency in the use of natural resources. As a result of this study, I have found that the potential development of a self-organizing homemade food market in neighborhoods could improve the energy efficiency of the food system, allowing energy savings equal to 1.2 percent of the total energy used by a city’s food system. To support the development of such a market, tools used in AI applications such as sentiment analysis and the Blockchain, can play a significant role in their ability to anticipate market needs, and provide a secure, transparent, and efficient transaction platforms. Therefore, these AI tools can have a significant impact in reducing the need of trustees such as banks and companies that provide access to peer-to-peer service platforms like Uber and Airbnb.
Large-scale network analytics
(2011-08) Song, Han Hee, 1978-; Zhang, Yin, doctor of computer science
Scalable and accurate analysis of networks is essential to a wide variety of existing and emerging network systems. Specifically, network measurement and analysis helps to understand networks, improve existing services, and enable new data-mining applications. To support various services and applications in large-scale networks, network analytics must address the following challenges: (i) how to conduct scalable analysis in networks with a large number of nodes and links, (ii) how to flexibly accommodate various objectives from different administrative tasks, (iii) and how to cope with the dynamic changes in the networks. This dissertation presents novel path analysis schemes that effectively address the above challenges in analyzing pair-wise relationships among networked entities. In doing so, we make the following three major contributions to large-scale IP networks, social networks, and application service networks. For IP networks, we propose an accurate and flexible framework for path property monitoring. Analyzing the performance side of paths between pairs of nodes, our framework incorporates approaches that perform exact reconstruction of path properties as well as approximate reconstruction. Our framework is highly scalable to design measurement experiments that span thousands of routers and end hosts. It is also flexible to accommodate a variety of design requirements. For social networks, we present scalable and accurate graph embedding schemes. Aimed at analyzing the pair-wise relationships of social network users, we present three dimensionality reduction schemes leveraging matrix factorization, count-min sketch, and graph clustering paired with spectral graph embedding. As concrete applications showing the practical value of our schemes, we apply them to the important social analysis tasks of proximity estimation, missing link inference, and link prediction. The results clearly demonstrate the accuracy, scalability, and flexibility of our schemes for analyzing social networks with millions of nodes and tens of millions of links. For application service networks, we provide a proactive service quality assessment scheme. Analyzing the relationship between the satisfaction level of subscribers of an IPTV service and network performance indicators, our proposed scheme proactively (i.e., detect issues before IPTV subscribers complain) assesses user-perceived service quality using performance metrics collected from the network. From our evaluation using network data collected from a commercial IPTV service provider, we show that our scheme is able to predict 60% of the service problems that are complained by customers with only 0.1% of false positives.
Learning to write in (networked) public: children and the delivery of writing online
(2014-12) Roach, Audra Katherine; Bomer, Randy; Hoffman, Jim; Maloch, Beth; Schallert, Diane; Hodgson, Justin
This investigation explored how three children (together with parents) developed networked publics that were diverse, well-connected, and powerful in the world. It was framed in response to calls in the field to better understand the new literacies young writers develop online and outside of school, and to increase literacy educators’ attention to the role of public audiences in writing and how writing is circulated. Performative case study methodology, ethnographic methods, and digital methods were employed to track and describe the online networks of three children (ages 11-13). These focal children were actively involved with their parents in social media, and had developed widespread networks with shared interests in children’s books and book reviews (Case 1), baseball (Case 2), and helping the homeless (Case 3). The children’s online networks were conceptualized as networked publics, drawing on Warner’s (2002) notion of publics as ongoing discursive relations among strangers, and on Actor-Network Theory’s notion of networks as assemblages of diverse interests that mobilize toward a common goal (Callon, 1986) and that develop stability in relation to ongoing circulations of texts (Latour, 1986; Spinuzzi, 2008). Research questions were framed broadly around the rhetorical canon of delivery [now digital delivery (Porter, 2009)], and were concerned with how writers distributed texts online, how those texts circulated, how the networked publics become more stable and powerful, and what instabilities children and parents had to negotiate in order to accomplish all of this. Data sources included interviews with 15 children and 28 adults, and fieldnotes observations of approximately 1,700 screen-captured webpages and other online artifacts. Findings showed that the young writers and their parents initiated and sustained networked publics through distribution practices that were oriented toward building trust; their texts displayed: interest, appreciation, reliability, service, credibility, and responsiveness. Both grassroots and commercial entities circulated texts in these networks, as they were sources of the ongoing renewal these different groups all needed in order to thrive. Sources of instability included conflicts over standards of writing quality, matters of profit, and the constancy of the demand to generate new interest and writing online. Children and their parents responded to these instabilities by welcoming and negotiating heterogeneous perspectives and partnerships. Implications of the study call for further research and teaching about the art of networked public discourse and digital delivery.
Microscale modeling of layered fibrous networks with applications to biomaterials for tissue engineering
(2015-08) Carleton, James Brian; Rodin, G. J. (Gregory J.); Sacks, Michael S.; Gonzalez, Oscar; van de Geijn, Robert; Mear, Mark
Many important biomaterials are composed of multiple layers of networked fibers. A prime example is in the field of tissue engineering, in which damaged or diseased native tissues are replaced by artificial tissues that are grown on fibrous polymer networks. For load bearing tissues, it is critical that the mechanical behavior of the engineered tissue be similar to the behavior of the native tissue that it will replace. In the case of soft tissues such as heart valves, the macroscale mechanical behavior is highly anisotropic and nonlinear. This behavior is a result of complex deformations of the collagen and elastin fibers that form the extracellular matrix (ECM). The microstructure of engineered tissues must be properly designed to reproduce this unique macroscopic behavior. While there is a growing interest in modeling and simulation of the mechanical response of this class of biomaterials, a theoretical foundation for such simulations has yet to be firmly established. This work introduces a method for modeling materials that have a layered, fibrous network microstructure. Methods for characterizing the complex network geometry are first established. Then an algorithm is developed for generating realistic network geometry that is a good representation of electrospun tissue scaffolds, which serve as the primary synthetic structure on which engineered tissues are grown. The level of fidelity to the real geometry is a significant improvement on previous representations. This improvement is important, since the scaffold geometry has a strong influence over the macroscopic mechanical behavior of the tissue, cell proliferation and attachment, nutrient and waste flows, and extracellular matrix (ECM) generation. Because of the importance of scaffolds in tissue formation and function, this work focuses on characterizing scaffold network geometry and elucidating the impact of geometry on macroscale mechanics. Simulation plays an important role in developing a detailed understanding of scaffold mechanics. In this work, Cosserat rod theory is used to model individual fibers, which are connected to form a network that is treated as a representative volume element (RVE) of the material. The continuum theory is the basis for a finite element discretization. The nonlinear equations are solved using Newton's method in a parallel implementation that is capable of accurately capturing the large, three-dimensional fiber rotations and large fiber stretches that result from the large macroscopic deformations experienced by these biomaterials in their natural environment. Comparisons of simulation results with existing analytical models of soft tissues show that these models can predict the behavior of scaffold networks with reasonable accuracy, despite the significant differences between soft tissue and scaffold network microstructural geometry. The simulations also reveal how macroscale loading is related to the microscale fiber deformations and the load distribution among the fibers. The effects of different characteristics of the microstructural geometry on macroscopic behavior are explored, and the implications for the design of scaffolds that produce the desired macroscopic behavior are discussed. Overall, the improved modeling of electrospun scaffolds presented in this work is an important step toward designing more functional engineered tissues.
MobiShare : mobile computing with no strings attached
(2013-12) Castillo, Jason Moses; Julien, Christine, D. Sc.
In today’s world, technology is growing at a fast rate compared at other times. Sales have increased in the smart phone market, which has created new opportunities in pervasive computing. In pervasive computing, nodes enter and leave a network at any time. Within the network, nodes can transfer data to other nodes. The information is not retained in any static location such as a server. The mobile infrastructure requires a way to handle all the information in a dynamic way. The use of a centralized server in a mobile environment creates deterioration in the performance of obtaining information. The main goal of this paper is to provide data persistence using a “substrate” that is inherently not persistent. The data will be stored within the network for availability to all users. Saving data within a network would provide a means to obtain any type of information without relying on the source of where the data came from in the network. Users would also be able to continue downloading where they left off when they return to the network. Consider an environment where people can share music or books. For example, say that John Doe was searching for a particular song to download and in the network Jane has the song that was requested. John decides to download the song without knowing that it is from Jane. Then John decides to leave the network and the download stops. Whenever John rejoins the network the download of his song will continue where he left off, and his ability to access the information will not depend whether or not Jane is present in the network. John may retrieve the file from any other user who has the exact same file. The requested information that the user queries in a search engine will be stored as a metadata within the network, either by other nodes or a temporary server. This allows data to be obtained without relying on the "main user" or creator of the data to be present in the network. The users would also be able to retrieve the data at multiple times.
Network strengthening for policy influencing : a case study of Kenya’s Africa Adaptation Programme (AAP) of the United Nations Development Programme
(2011-12) Nkaw, John; Weaver, Catherine, 1971-; Busby, Jushua
As researchers provide compelling evidence pointing to climate change, governments and civil society actors are getting stimulated to act and reverse the negative impacts of extreme climate change. The impact of climate change on Kenya is profound and staggering. It is estimated that Kenya’s landmass is 582,350 km2, of which only 17% is arable, with 83% consisting of semi-arid and arid land. Climate change and human activities are resulting in desertification and increasing total semi-arid and arid land. Researchers further estimate that 17% of Mombasa or 4600 hectares of the region’s land area will be submerged as a result of sea-level rise. This situation demands policy actions to combat the situation. As developing countries wade into combating climate change, the government of Kenya is implementing far reaching polices to fight climate change including its 2006 water quality regulation and 2009 regulation of wetlands, riverbanks, lakeshore and sea management regulations of 2009. In addition, development partners such as the UNDP and civil society actors working on climate change have played a critical role complementing government policy actions. Working through the Africa’s Adaptation Programme (AAP), civil society organizations (CSOs) are participating in agenda setting, and increasing awareness that promote climate change adaptation through civic engagement. Civic engagement serves as an important tool for nongovernmental organizations (NGOs) to promote a more effective response to the hazardous effects of extreme climate change. Despite this, researchers and policy analyst argue that civil societies work within the environmental sector is not based on rigorous research, their actions are uncoordinated, and outcomes are poorly communicated. As a focal point, this report examined how CSOs organize around key policy issues and work through the AAP to set the agenda and influence climate change policymaking in Kenya. The study is based largely on an evaluation of secondary data sources including websites, Programme documents and academic articles. I also benefited from a summer internship at UNDP offices in Nairobi in 2010. The study explored how AAP is professionalizing and how that increases its leverage and strengthens NGOs to actively participate in policy influencing. The study summarizes scattered pieces of information into one report to enhance the AAP’s database building efforts. Finally, this serves as resource for CSOs policy engagement in Kenya and beyond. Overall, the report reveals that the AAP is bridging ties between CSOs working within the climate change sector by bringing them under one umbrella. This social bonding behavior serves as social capital to influence policy. However to increase leverage for effective policy engagement, the AAP needs to incrementally apply rigorous evidenced based research to generate more compelling information that transforms policies. It further suggests commercializing clean energy technologies by charging affordable rates for deploying such infrastructure to households. Finally, using policy entrepreneurs can dramatically improve policy advocacy in Kenya.
Networking abstractions for GPU programs
(2015-08) Kim, Sangman; Witchel, Emmett; Alvisi, Lorenzo; Ford, Bryan; Pingali, Keshav; Porter, Donald E.; Silberstein, Mark
Graphics Processing Units (GPUs) are becoming major general-purpose computing hardware for high-performance parallel computation. Despite their general computation capability and impressive performance, GPUs still lack important operating system (OS) services like networking, which makes building networking services or distributed systems on GPUs challenging. This thesis presents GPUnet, a native GPU networking layer that provides a socket abstraction and high-level networking APIs for GPU programs. GPUnet abstracts complicated coordination of processors and network interfaces from GPU applications, and streamlines the development of server applications on GPUs. We develop several applications that harness the benefit of GPUnet: our matrix multiplication server with GPUnet's performance matches or surpasses the performance of the server without GPUnet, with only 24-43% of lines of code. We also show the scalability of in-GPU-memory MapReduce (GimMR) applications across multiple GPUs. Its word count and K-means workloads can scale to four GPUs with speedups of 2.9-3.5x over one GPU. GPUnet addresses three key challenges: massive parallelism of GPU programs, memory copy overhead between CPU memory and GPU memory, and slow single-threaded performance of GPUs. To better support massively parallel GPU programs, the networking API invocations from multiple threads at the same point in a data-parallel code are coalesced. Direct communication between GPUs and the network devices reduces the copy overhead, and, to minimize the amount of time spent in the single-threaded operation, the control-intensive networking operations are offloaded to the network device.
Policy network and content analysis : applications in water resources management and science
(2017-07-18) Wolf, Emery Charles; Pierce, Suzanne Alise
This study extends previous work using the state water plans from 1961-2017 with the most recent 2016 regional plan submissions from the Texas Water Development Board, to implement and evaluate a topic analysis methodology. The approach uses statistical analysis of the collection of text documents or corpus to evaluate. Topic Modeling is a systematic approach for analyzing the relationships, usage frequency of words and communities of words to extract themes, concepts, and informational meaning from a selected corpus. This research documents methods for content analysis that can be used on state water plans, as well as other environmental science and policy documents. For this study, nearly 19,658 pages of text from the state and regional water plans for Texas were analyzed. Unsurprisingly, results indicate that “water” is the central common theme connecting all topics. Early results identified a set of primary topics that are shared throughout all regions including planning, strategy, and groundwater. Interestingly, themes varied from west to east reflecting the gradient of arid to humid climates respectively. In the West, themes indicate that regional water planning groups focus more heavily on irrigation and wells for agriculture, while in the East the focus tends to be for municipal uses and surface water strategies, such as reservoirs and infrastructure. This thematic pattern also aligns with the population distribution of Texas, with larger numbers of people in the east, and much less dense populations in the west. Analyses of the state water plans over time illustrate that topics related to drought, planning, and water needs have increased over the period under study. Network statistics reveal that the largest change between state water plans occurred between the 1961 and 1968 plans. Topic analysis methodologies provide an accessible and systematic approach to evaluate the context of water planning, management, and policy across the state. The approach may provide a mechanism for linking quantitative science knowledge about water resources in the state with the qualitative planning and policy perspectives used to manage these critical resources.
Resource-constrained, scalable learning
(2015-08) Mitliagkas, Ioannis; Vishwanath, Sriram; Caramanis, Constantine; Dimakis, Alex; Sanghavi, Sujay; Ravikumar, Pradeep
Our unprecedented capacity for data generation and acquisition often reaches the limits of our data storage capabilities. Situations when data are generated faster or at a greater volume than can be stored demand a streaming approach. Memory is an even more valuable resource. Algorithms that use more memory than necessary can pose bottlenecks when processing high-dimensional data and the need for memory-efficient algorithms is especially stressed in the streaming setting. Finally, network along with storage, emerge as the critical bottlenecks in the context of distributed computation. These computational constraints spell out a demand for efficient tools that guarantee a solution in the face of limited resources, even when the data is very noisy or highly incomplete. For the first part of this dissertation, we present our work on streaming, memory-limited Principal Component Analysis (PCA). Therein, we give the first convergence guarantees for an algorithm that solves PCA in the single-pass streaming setting. Then, we discuss the distinct challenges that arise when the received samples are overwhelmingly incomplete and present an algorithm and analysis that deals with this issue. Finally, we give a set of extensive experiment results that showcase the practical merits of our algorithm over the state of the art. The need for heavy network communication arises as the bottleneck when dealing with cluster computation. In that paradigm, a set of worker nodes are connected over the network to produce a cluster with improved computational and storage capacities. This comes with an increased need for communication across the network. In the last part of this work, we consider the problem of PageRank on graph engines. Therein, we make changes to GraphLab, a state-of-the-art platform for distributed graph computation, in a way that leads to a 7x-10x speedup for certain PageRank approximation tasks. Accompanying analysis supports the behaviour we see in our experiments.
TASNIC : a flexible TCP offload with programmable SmartNICs
(2021-05-07) Shashidhara, Rajath; Peter, Simon, Ph. D.
The CPU overhead of TCP packet processing is increasingly prohibitive. Kernel-bypass stacks and existing hardware offloads are not enough, causing operational issues at scale and cannot keep up with network protocol evolution. We take advantage of programmable SmartNICs to build TASNIC, a flexible, yet high-performance TCP offload engine (TOE) in software. TASNIC eliminates almost all host CPU packet processing, but retains compatibility with POSIX sockets and places no demands on data center networks. TASNIC allows complete customization of transport logic, providing performance and flexibility. TCP offload to SmartNICs is challenging. SmartNICs are geared towards massively parallel stateless offloads, while TCP is a complex stateful protocol, sensitive to packet reordering. TASNIC leverages fast-path processing of common TCP code paths, fine-grained parallelization of the TCP data-path, and near-memory computing for high performance, while remaining flexible via a modular design. We compare TASNIC to Linux, the TAS TCP accelerator, and the Chelsio Terminator TOE. We find that Memcached scales up to 38% better on TASNIC versus TAS, while saving up to 80% host CPU cycles versus Chelsio. For 64B RPCs, TASNIC cuts 99th-percentile latency to 42% and provides 70% higher throughput versus TAS, and an order of magnitude higher throughput under packet loss than Chelsio. TASNIC interoperates well with other TCP stacks and is easily extensible.
Web news in China : a new hierarchy of centrality? : an analysis of the linking pattern of China’s online news network
(2010-05) Chen, Xin, 1977 Aug. 2-; Lasorsa, Dominic L.; Poindexter, Paula M.; Straubhaar, Joseph; Chyi, Hsiang I.; Alves, Rosental C.
The present study explored three questions: What is the linking pattern of China’s cyber news space? What are the factors that contribute to this pattern? And what is the distribution of links in real geographic places? The concept of the cyber news space refers to the globally connected networks of online news production. It is a tool to understand the spatial distribution of online news production and the map of the world as presented in the media. This study is a content analysis of news Webpages from China’s four leading commercial portals. It sampled about 900 news Webpages during the spring of 2008. China’s commercial portals are news aggregators and distributors. They are the gatekeepers of China’s cyber news space. On their news Webpages there is one hyperlink that leads to the original publisher of the story. These links provide a clue of how news organizations were connected online. The content analysis coded these links and other information such as media type, production sites and locations of stories. This study found that the there was a pattern of concentration in terms of the distribution of links among online news organizations. A multiple regression model was used to test the factors that may contribute to this pattern. It was found that geographic location of news organizations was such a factor. The more central a news organization was located, the more links it attracted from the portals. In addition, this study also analyzed the distribution of links among difference provinces (or province level administrations) of China. It found that Beijing, Chongqing, Guangdong, Jilin, and Shanghai are hubs, while more remote provinces, such as Xinjiang, and Guizhou were largely bypassed.