With the adoption of cloud services and the reliability and resiliency it offers, enterprises are eager to understand how many of their legacy applications can be containerized. We propose Application Containerization advisor (ACA), a framework that provides a containerization advisory for legacy applications. Given an application description in terms of its technical components, ACA proposes a multi-step process that standardizes the raw inputs and curates technology stack into various components, detects missing components and finally recommends the best possible containerization approach.

To support IT change automation in cloud-native deployments, we propose to discover change actions from IT change requests. Traditional approaches to discover actions rely on a pre-established set of actions. However in practice, the catalog dictionaries rarely cover a sufficient portion of the IT change requests, resulting in missed automation opportunities. To this end, supervised and semi-supervised approaches have been proposed to detect change actions, but fall short in covering new action discovery as the IT environment evolves. We propose an ensemble technique of parametric and non-parametric grammar-based methods to discover change actions from IT change requests. We evaluate our approach on an IT dataset and find our approach provides significant coverage of actions compared to baseline approaches.

When optimizing software for the cloud, monolithic applications need to be partitioned into many smaller microservices. While many tools have been proposed for this task, we warn that the evaluation of those approaches has been incomplete; e.g. minimal prior exploration of hyperparameter optimization. Using a set of open source Java EE applications, we show here that (a) such optimization can significantly improve microservice partitioning; and that (b) an open issue for future work is how to find which optimizer works best for different problems. To facilitate that future work, see https://github.com/yrahul3910/ase-tuned-mono2micro for a reproduction package for this research.

In migrating production workloads to cloud, enterprises often face the daunting task of evolving monolithic applications toward a microservice architecture. At IBM, we developed a tool called Mono2Micro to assist with this challenging task. Mono2Micro performs spatio-temporal decomposition, leveraging well-defined business use cases and runtime call relations to create functionally cohesive partitioning of application classes. Our preliminary evaluation of Mono2Micro showed promising results.

Mono2Micro is an AI-based toolchain that provides recommendations for decomposing legacy web applications into microservice partitions. Mono2Micro consists of a set of tools that collect static and runtime information from a monolithic application and process the information using an AI-based technique to generate recommendations for partitioning the application classes. Each partition represents a candidate microservice or a grouping of classes with similar business functionalities. Mono2Micro takes a temporo-spatial clustering approach to compute meaningful and explainable partitions. It generates two types of partition recommendations. First, it computes business-logic-seams-based partitions that represent a desired encapsulation of business functionalities. However, such a recommendation may cut across data dependencies between classes, accommodating which could require significant application updates. To address this, Mono2Micro computes natural-seams-based partitions, which respect data dependencies. We describe the set of tools that comprise Mono2Micro and illustrate them using a well-known open-source JEE application.

Existing approaches to recommend services using natural language queries are supervised or unsupervised. Supervised approaches rely on a dataset with natural language queries annotated with categorizing labels. As the annotation process is manual and requires deep domain knowledge, these approaches are not readily applicable on new datasets. On the other hand, unsupervised approaches overcome the limitation. To date, unsupervised approaches are primarily based on matching keywords, entity relationships, topics and clusters. Keywords and entity relationships ignore the semantic similarity between a query and services. Topics and clusters capture the semantic similarity, but rely on mashups that explicitly capture relationships between services. Again, for new services, the information are not readily available. We propose NL2API, a framework that relies solely on service descriptions for recommending services. NL2API has the benefit of being immediately applicable as a bootstrap recommender for new datasets. To capture relationships among services, NL2API provides different approaches to construct communities where a community represents an abstraction over a group of services. Based on the communities and users' queries, NL2API applies a query matching approach to recommend top-k services. We evaluate NL2API on datasets collected from Programmable Web and API Harmony. Our evaluation shows that for sizable datasets such as Programmable Web NL2API outperforms baseline approaches.

Presented BlueWall: Software Defined Network Management in Hybrid Enterprise Cloud Environments for managing firewall request when servers (or services) are created (via APIs). Demonstrated hybrid network management design and self-service capabilities. Discussed challenges arising in network management in the hybrid enterprise cloud environments.

In the last few years, changing infrastructure and business requirements are forcing enterprises to rethink their networks. Enterprises look for network infrastructures that increase network efficiency, flexibility, and cost reduction. At the same time, the emergence of Cloud and mobile in enterprise networks has introduced tremendous variability in enterprise traffic patterns at the edge. This highly mobile and dynamic traffic presents a need for dynamic capacity management and adaptive traffic steering and appeals for new infrastructures and management solutions. In this context, passive optical networks (PON) have gained attention in the last few years as a promising solution for enterprise networks, as it can offer efficiency, security, and cost reduction. However, network management in PON is not yet automated and needs humain intervention. As such, capabilities for dynamic and adaptive PON are necessary. In this paper, we present a joint solution for PON capacity management both in deployment and in operation, as to maximize peak load tolerance by dynamically allocating capacity to fit varying and migratory traffic loads. To this end, we developed the novel approaches of capacity pool based deployment and dynamic traffic steering in PON. Compared with traditional edge network design, our approach significantly reduces the need for capacity over-provisioning. Compared with generic PON networks, our approach enables dynamic traffic steering through software-defined control. We implemented our design on a production grade PON testbed, and the results demonstrate the feasibility and flexibility of our approach.

With the increased popularity of mobile data applications, Wi-Fi power consumption on smartphones is now a significant portion of mobile energy expenditure. In their efforts to develop more energy efficient applications, the applications developers use energy estimation tools as a benchmark. Although hardware based power meter has high estimation accuracy, it is cumbersome to operate as it relies on physically attaching wires to the battery, and moreover the price of hardware based power meter is expensive. Therefore software based power estimation tools such as PowerTutor are popular. In our prior research using PowerTutor as power meter, we discovered that PowerTutor has large Wi-Fi power estimation error (over 1000%) on post 2012 phones. In this work, we propose a new FSM-based Wi-Fi power model based on IEEE 802.11 communication patterns and Wi-Fi hardware configuration with significantly increased power estimation accuracy. We designed and implemented our power model as an estimation tool called PowerGuide. Our experiments with PowerGuide in field operations showed that PowerGuide can achieve an average estimation accuracy of 86% compared to hardware power meters even with moderate polling period.

Recent advances in the manufacturing and commercialization of passive optical components are now extending the capabilities of fiber to edge and campus networks. This paper presents a comparison between Passive Optical LAN (POL) and copper-based LAN solution, and demonstrate the benefits of PON such as reduced infrastructure footprint and cost, reduced power requirements, future-proof bandwidth, greener infrastructure, safer, higher security and higher reliability.

Wi-Fi is a popular wireless communication technology for smart devices such as smartphones, however, Wi-Fi related energy consumption in smartphones contributes to a significant portion of its energy expenditure. Hence it is important to carefully analyze and profile Wi-Fi energy expenditure in order to improve mobile applications' energy efficiency. Although hardware based power meters provide very accurate power measurement result, it is infeasible to expect mobile application developers to rely on power meters. As alternative solutions, software based power estimation tools are popular. Most of the existing power estimation solutions to date do not provide accurate estimation result, as we disclose in this paper by analyzing the popular PowerTutor. Therefore, we propose a new Wi-Fi power model by taking into account important IEEE 802.11 communication patterns as well as Wi-Fi hardware settings in a way that increases the accuracy of power estimation. We design and implement the proposed power model as a smartphone application and deploy it into a real device for validation. The evaluation results show that our solution can achieve an estimation accuracy up to 86% compared to hardware power meters, and is much more accurate than PowerTutor.

Data communication is a significant component of smart device energy consumption today, and is likely to increase with the rising connectivity demands of today's mobile applications. Wi-Fi is a major supporting technology and as such, efficient energy management solutions are much needed. In this paper, we detail the Wi-Fi energy consumption characteristics and propose two energy saving schemes: packing and alignment, implementable at the application layer. Our investigations also reveal that these schemes are highly reliant on the network metric - available bandwidth, which is not readily obtainable on smartphones and existing wired solutions are ill-suited due to the dynamics of the wireless environment, as well as the energy constraint of mobile devices. Therefore, we propose a new energy efficient bandwidth measurement tool called BreezChirp. As validation, we implemented both BreezChirp and the Wi-Fi management schemes on modern Andriod smartphones and evaluated their performance through field experiments.

In the presence of multiple overlays and underlays, the emerging global network behavior is the result of interactions of self-serving overlay routing decisions and independent underlay management actions. It is crucial for network operators, service, and content providers to have a good grasp of the underlying principles in order to better design and manage current and future networks and services. In this paper, we describe special game scenarios wherein the interaction of noncooperative overlays and underlays in multidomain networks can result in an operable global configuration in linear time and the overall convergence is polynomial in the unweighed case. For weighted games, we find that weighted Shapley potential can achieve linear time convergence to an operable state. Furthermore, we analyze the interaction of overlays and underlays as a two-stage congestion game and recommend simple operational guidelines to ensure global stability. We further explore the use of Shapley value as an enabler of mutual cooperation in an otherwise competitive environment. Our simulation results confirm our findings and demonstrate its effectiveness in general networks.

In managing multimedia services, it is important to understand how network performance affects user experience. The model presented in this paper aims to estimate user perception of video quality based on defect events, which are automatically classified by machine learning techniques. The underlying principle of our model is that human experience is event-based and there is a strong correlation between defective events and user MOS. Through experiments, we show that our model can detect different types of defect events with good accuracy even under small data set, and we find that indeed different defect event types affect user experience with different sensitivity.

Vast majority of the services running on the smart phone today are networked in that significant amount of communication is required. Smart phone energy expenditure due to Wi-Fi communications constitutes significant portion of the battery discharge. In this paper, we first investigate the key factors influencing Wi-Fi energy consumption, and propose three energy management schemes: 1) Dynamic control of Wi-Fi on/off interface; 2) improve communication efficiency via application packing; and 3) elongation of Wi-Fi Power Save Mode (PSM) via application alignment under mixed application workload. We also design and test our solution as a device-side application utilizing general system process scheduling and network firewall techniques. As the result, our solution is easy to deploy, applicable to most mobile devices, and we explicitly tackle the challenging case of download management. Through extensive experimentation and solution prototyping, we show the effectiveness of device side Wi-Fi energy management and the importance of considering application characteristics.

Multimedia services have become a dominant part of the network and content provider's service portfolio. A formal modeling methodology for video quality assessment not only affords the providers a clear cause-effect management view of their services, but also helps to guide management and planning operations. We examine the relation between network performance and perceived video quality through VIDAR, a comprehensive VIDeo quality Analyzer in Real-time. VIDAR links network performance to objective frame quality (eSSIM), and modify eSSIM values by subjective filters. Through experiments, we show that VIDAR helps providers to better understand and assess the impact of the network performance on perceived video quality.

In this paper, we present demand-side energy management under real-time demand-response pricing as a task scheduling problem which is NP-hard. Using minmax as the objective, we show that the schedule produced by our minMax scheduling algorithm has a number of salient advantages: significant peak-shaving, cost reduction, and risk-aversion for the consumers. We prove that our algorithm finds near-optimal solutions and our simulation study show that the actual performance is better than the worst-case bound. The algorithm is simple to implement and efficient at the scale of large enterprises.

Network virtualization environment (VNE) affords great business flexibility to the customers and the providers as multiple providers can jointly support a customer's virtual network. Under the current network model, a group of Infrastructure Providers (InPs) peer with each other to provide a packaged deal. Such a business arrangement is not customer-driven, does not promote fair market competition and does not ensure cost minimization. Furthermore, the on-demand nature of virtual networks requires efficient and automated service negotiation and contracting. In this paper, we present V-Mart. To the InPs, V-Mart offers an environment to participate in a faithful and fair competition over the VN resources; and to the SPs, it offers a customer-driven virtual resource partitioning and contracting engine. V-Mart uses a two-stage Vickrey auction model that is strategy-proof, flexible to diverse InP pricing models, and functions over heterogenous multi-commodity market that characterizes the NVE. Through analysis and simulation we show the flexibility and effectiveness of V-Mart.

In recent years application-level networking (e.g. overlay networks, P2P networks) has become key enabling architecture for supporting distributed services on the Internet. We consider overlays to have delay and cost requirements and thus there exists an optimal distribution of the overlay traffic (the service engineering problem) across multi-domain underlay networks such that these requirements can be satisfied while achieving fairness among the overlays. To this end, we formulate the problem as a convex optimization problem and propose a distributed solution. We show the efficiency and effectiveness of our approach through simulation studies.

With the increased network size and diversity, and the proliferation in applications and services, the network service providers are faced with a flood of information from many levels of their network and service operations, often in uncorrelated forms. At the same time, the focus of network services is shifting from managing networks to managing services and customers. CyberPlanner is a comprehensive toolkit for the collection, correlation and filtering of metrics at all levels of a network service providerpsilas business. In this paper, we present its concept, modeling methodologies and implementation. Experience gained from CyberPlanner testing and operations in both academic and industrial settings demonstrates its effectiveness and potential in assisting many key provider operations such as customer-oriented trouble diagnostics, network dimensioning and upgrading, service planning and business revenue forecasting.

Next generation networks are envisioned to support dynamic and customizable service compositions at Internet scale. To facilitate the communication between distributed software components, on-demand and QoS-aware network service composition across large scale networks emerges as a key research challenge. This paper presents a fast QoS-aware service composition algorithm for selecting a set of interconnected domains with specific service classes. We further show how such algorithm can be used to support network adaptation and service mobility. In simulation studies performed on large scale networks, the algorithm exhibits very high probability of finding the optimal solution within short execution time. In addition, we present a distributed service composition framework utilizing this algorithm.

Advents in network technology and distributed system design have propelled network communication service beyond best effort data delivery. With the rising complexity of network infrastructures and the need for on-demand provisioning operations, a high degree of self-sufficiency and automation is required in the network service infrastructure. Guided by the autonomic communication principle, this paper first presents an autonomic service provisioning framework for establishing quality-of-service (QoS)-assured end-to-end communication paths across administratively independent domains. Through graph abstraction, we show that the domain composition and adaptation problem could be reduced to the classic k-multiconstrained optimal path (MCOP) problem. In analyzing existing k-MCOP solutions, we show their inefficiencies when applied to the service provisioning context and establish a number of new domain composition and adaptation algorithms. These new algorithms are designed for the self-configuration, self-optimization, and self-adaptation of end-to-end network communications and can provide hard QoS guarantees over domains with relative QoS differentiations. Through in-depth experimentations, we compare the performance of our algorithms with classic k-MCOP solutions and demonstrate the effectiveness of our approach.

With the ever increasing demand for network resources, network operators and Internet service providers are under constant pressure to accommodate more network bandwidth and offer better service quality via periodic network upgrades, Given a budget constraint, a sound network upgrade decision should maximize investment benefit which is contingent on the degree of customer satisfaction. This paper presents a customer-centric approach in making network upgrade decisions, where customer satisfaction is the key evaluation criterion. Network performance is related to customer's perceived service quality and component upgrades are assessed based on their profitability. As demonstrated using a case scenario, our approach results in effective upgrade decisions that enhance service quality, improve customer satisfaction, and maximize revenue.