HTTP Sessions Research Articles

Method of horizontal pod scaling in Kubernetes to omit overregulation

This paper describes the method of omitting the over-regulation effect that occurs under certain conditions by horizontal pod autoscaling microservices in the container application orchestration system Kubernetes. The effect was initially observed only for long-term HTTP WebSocket sessions, where it led to excessive use of computing resources, which reduced the efficiency of IT infrastructure management, and caused service failure. It was found that the overregulation effect is reproduced not only for connections with long-term HTTP sessions, such as HTTP WebSocket, but also for shorter-term REST HTTP sessions in case of increased delay in the metric collection cycle used for horizontal pod autoscaling. It is assumed that this effect happens due to the approach of implementing horizontal scaling controllers similar to the principles of proportional regulators in systems with negative feedback from the theory of automation and control. It is proposed to extend one of the methods used for optimizing the proportional controller to the problem consisting of reducing the time delay between scaling metrics collecting and upscale applied by the controller in Kubernetes. The applied method demonstrated its effectiveness, therefore, within the same methodology, an experiment was conducted on using the proportional-integral-differential controller for automatic horizontal scaling of pods. The results obtained showed why the proportional-integral-differential controller is not widespread among the overviewed Kubernetes solutions for horizontal automatic scaling. An assumption was made about the limitations of studying the downscaling process in Kubernetes due to the need to consider the quality of service when stopping pods and the need to collect indicator metrics using quality-of-service object management tools such as ISTIO.

Business process model with “RandomForest” algorithm

Currently, the areas of application of machine learning are multifaceted: artificial intelligence, financial applications, bioinformatics, intellectual games, speech and text recognition, computer language processing, medical diagnostics, technical diagnostics, text search and rubrication. Machine learning is an area of scientific knowledge associated with learning-capable algorithms. The use of machine learning methods is explained by the fact that for most intelligent complex tasks (for example, speech recognition, etc.) it is almost impossible to develop an obvious algorithm for their solution. However, you can teach a computer to learn how to solve such problems. In our article, we propose a model based on the machine learning algorithm “RandomForest”, which allows one to recognize bots by HTTP sessions. The chosen algorithm provides many advantages: non-iterative learning; high quality of the resulting models (comparable to neural networks and ensembles of neural networks); a small number of adjustable parameters. It works well with missing data (retains good accuracy); internal assessment of the generalizability of the model; able to work with raw data without preprocessing. The algorithm was trained on a dataset of more than 5000 sessions. The prospects of this direction are obvious, since robotic traffic accounts for more than 40% of the total Internet traffic.

Detecting and characterising of mobile advertisement network traffic using graph modelling

Many 'free' applications are provided for Android. These include advertisement (ad) modules manage ad services and track user sensitive behaviour. These sometimes lead violations of privacy. We analysed 797 of 1,188 applications included 45 known ad modules and found characteristic ad network traffic patterns. In order to accurately differentiate traffic between ad modules and valid application, we propose a novel method based on the distance between traffic graphs mapping the relationships between HTTP sessions. Using this method, we can detect ad modules' traffic by comparing session graphs with known ad graphs. In evaluation, we generated 20,903 graphs from applications traffic includes 4,698 known ad graphs, manually identified 2,000 ad graphs, and 2,000 standard application graphs. We also evaluated graph screening for detection accuracy. Our approach showed 76% detection rate for known ad graphs, 96% detection rate for manually classified ad graphs, and under 10% false positive rate for standard graphs.

Passive Smart Phone Indentification and Tracking with Application Set Fingerprints

Current smart phone users can be identified and tracked by fingerprint techniques. Fingerprint identification techniques can be used for legitimate purposes such as network management and traffic control to avoid excessive congestion. This paper proposes a user identification and tracking technique specific to smart phone users for supervised network management. This paper proposes the application set fingerprint, which is a simple set of User-Agent request-header fields in HTTP sessions. The application set fingerprint has three advantages, fully-passive fingerprint generation, potential of user trackability, and fingerprinting considering the users' privacy. The results show that the application set fingerprint is practically effective and network operators can use it for tracking smart phone users with the purpose of efficient network management.

Survey on web spam detection

Search engines became a de facto place to start information acquisition on the Web. Though due to web spam phenomenon, search results are not always as good as desired. Moreover, spam evolves that makes the problem of providing high quality search even more challenging. Over the last decade research on adversarial information retrieval has gained a lot of interest both from academia and industry. In this paper we present a systematic review of web spam detection techniques with the focus on algorithms and underlying principles. We categorize all existing algorithms into three categories based on the type of information they use: content-based methods, link-based methods, and methods based on non-traditional data such as user behaviour, clicks, HTTP sessions. In turn, we perform a subcategorization of link-based category into five groups based on ideas and principles used: labels propagation, link pruning and reweighting, labels refinement, graph regularization, and featurebased. We also define the concept of web spam numerically and provide a brief survey on various spam forms. Finally, we summarize the observations and underlying principles applied for web spam detection.

Characterizing user navigation and interactions in online social networks

Understanding how users navigate and interact when they connect to social networking sites creates opportunities for better interface design, richer studies of social interactions, and improved design of content distribution systems. In this paper, we present an in-depth analysis of user workloads in online social networks. This study is based on detailed clickstream data, collected over a 12-day period, summarizing HTTP sessions of 37,024 users who accessed four popular social networks: Orkut, MySpace, Hi5, and LinkedIn. The data were collected from a social network aggregator website in Brazil, which enables users to connect to multiple social networks with a single authentication. Our analysis of the clickstream data reveals key features of the social network workloads, such as how frequently people connect to social networks and for how long, as well as the types and sequences of activities that users conduct on these sites. Additionally, we gather the social network topology of Orkut, so that we could analyze user interaction data in light of the social graph. Our data analysis suggests insights into how users interact with friends in Orkut, such as how frequently users visit their friends' and non-immediate friends' pages. Results show that browsing, which cannot be inferred from crawling publicly available data, accounts for 92% of all user activities. Consequently, compared to using only crawled data, silent interactions like browsing friends' pages increase the measured level of interaction among users. Additionally, we find that friends requesting content are often within close geographical proximity of the uploader. We also discuss a series of implications of our findings for efficient system and interface design as well as for advertisement placement in online social networks.

Preserving HTTP Sessions in Vehicular Environments

Wireless Internet in the in-vehicle environment is an evolving reality that reflects the gradual maturity of wireless technologies. Its complexity is reflected in the diversity of wireless technologies and dynamically changing network environments. The ability to adapt to the dynamics of such environments and to survive transient failures due to network handoffs are fundamentally important in failure-prone vehicular environments. In this paper we identify several new issues arising from network heterogeneity in vehicular environments and concentrate on designing and implementing a network-aware prototype system that supports HTTP session continuity in the presence of network volatility, with the emphasis on the following specifically tailored features: (1) automatic and transparent HTTP failure recovery, (2) network awareness and adaptation, (3) application-layer preemptive network handoff. Experimental results gathered from real application environments based on CDMA {\it 1xRTT} and IEEE 802 networks are presented and analyzed.

Congestion based resource sharing in multi-service networks

The internet traffic is increasingly composed of streams of packets with very diverse quality of service (QoS) requirements: from those containing standard, best-effort traffic, such as e-mail or http sessions, to internet telephony and high-quality video streaming demanding dedicated allocation of network capacity. However, while methods based on externality pricing exist to offer socially optimal prices to the groups of users demanding different grades of the best-effort service class, less attention has been devoted to the implementation of such pricing models to the improvement of network resources allocation in a multiservice setting (i.e., a network offering various grades of both guaranteed and best-effort level of service). This paper addresses the problem through an end-to-end approach that combines a congestion-based pricing model for resource sharing with a QoS network model, based on admission roles. We describe a measurement-based approach for QoS path pricing that results in improved resource allocation policies between QoS and best-effort traffic. We also present the results of a large-scale simulation study that shows how this pricing scheme increases both end user value and system throughput with low computational overhead.