Benjamin Satzger

Principles of Elastic Processes

Cloud computings success has made on-demand computing with a pay-as-you-go pricing model popular. However, cloud computings focus on resources and costs limits progress in realizing more flexible, adaptive processes. The authors introduce elastic processes, which are based on explicitly modeling resources, cost, and quality, and show how they improve on the state of the art.

Winds of Change: From Vendor Lock-In to the Meta Cloud

The emergence of yet more cloud offerings from a multitude of service providers calls for a meta cloud to smoothen the edges of the jagged cloud landscape. This meta cloud could solve the vendor lock-in problems that current public and hybrid cloud users face.

Esc: Towards an Elastic Stream Computing Platform for the Cloud

Today, most tools for processing big data are batch-oriented. However, many scenarios require continuous, online processing of data streams and events. We present ESC, a new stream computing engine. It is designed for computations with real-time demands, such as online data mining. It offers a simple programming model in which programs are specified by directed acyclic graphs (DAGs). The DAG defines the data flow of a program, vertices represent operations applied to the data. The data which are streaming through the graph are expressed as key/value pairs. ESC allows programmers to focus on the problem at hand and deals with distribution and fault tolerance. Furthermore, it is able to adapt to changing computational demands. In the cloud, ESC can dynamically attach and release machines to adjust the computational capacities to the current needs. This is crucial for stream computing since the amount of data fed into the system is not under the platforms control. We substantiate the concepts we propose in this paper with an evaluation based on a high-frequency trading scenario.

Application-level performance monitoring of cloud services based on the complex event processing paradigm

Monitoring of applications deployed to Infrastructure-as-a-Service clouds is still an open problem. In this paper, we discuss an approach based on the complex event processing paradigm, which allows application developers to specify and monitor high-level application performance metrics. We use the case of a Web 2.0 sentiment analysis application to illustrate the limitations we currently experience with regard to cloud monitoring, and show how our approach allows for more expressive definitions of monitored metrics. Furthermore, we indicate how the higher-level metrics produced by our approach can be used to increase application elasticity in an existing cloud middleware.

Auction-based crowdsourcing supporting skill management

Crowdsourcing is a promising approach for enterprises to maintain a flexible workforce that is able to solve parts of business processes formerly processed in-house. Companies perceive crowdsourcing as a concept that allows receiving solutions quickly and at little cost. Similar to cloud computing where computing power is provided on demand, the crowd promises a flexible on-demand workforce. However, businesses realize that these benefits entail a lack of quality control. The main difference compared to traditional approaches in business process execution is that tasks or activities cannot be directly assigned to employees but are posted to the crowdsourcing platform. Its members can choose deliberately which tasks to book and work on. In fact, crowdsourcing is heavily affected by the loose-coupling of workers to crowdsourcers and the dynamics of the environment. Hence, it remains a major challenge to guarantee high-quality processing of tasks within the prescribed time limit. A further obstacle for adoption of crowdsourcing in enterprises is the fact that it is hard to specify a fair monetary reward in advance. The concepts introduced in this work allow to smoothly integrate new workers, to keep them motivated, and to help them develop and improve skills needed in the system. We present a crowdsourcing marketplace that matches complex tasks, requiring multiple skills, to suitable workers. The key to ensuring high quality lies in skilled members whose capabilities can be estimated correctly. To that end, we present auction mechanisms that help to correctly estimate workers and to evolve skills that are needed in the system. Crowdsourcers do not need to predefine exact prices but only maximum prices they are willing to pay since the actual rewards for tasks are formed by supply and demand. Extensive experiments show that our approach leads to improved crowdsourcing, in most cases.

Cost-Efficient and Application SLA-Aware Client Side Request Scheduling in an Infrastructure-as-a-Service Cloud

Providers of applications deployed in an Infrastructure-as-a-Service cloud permanently face the decision of whether it is more cost-efficient to scale up(i.e., rent more resources from the cloud) or to delay incoming requests, even though doing so may lead to dissatisfied customers and broken service level agreements. This decision is further complicated by the fact that not all customers have the same agreements, and not all requests require the same amount of resources devoted to them. In this paper, we present an approach for optimally scheduling incoming requests to virtual computing resources in the cloud, so that the sum of payments for resources and loss incurred by service level agreement violations is minimized. We discuss our approach based on an illustrative use case. Furthermore, we present a numerical evaluation based on real-life request data, which shows that our agreement-aware algorithm improves upon earlier work, which does not take service level agreements into account.

A new adaptive accrual failure detector for dependable distributed systems

The detection of failures in distributed environments is a crucial part for developing dependable, robust, and self-healing systems. The contribution of this paper is a new failure detection algorithm that can be described as an adaptive accrual algorithm coupled with features to increase flexiblity and decrease computation costs. Furthermore our evaluation results show a very good detection quality in the case of message losses.

Stimulating skill evolution in market-based crowdsourcing

Crowdsourcing has emerged as an important paradigm in human problem-solving techniques on the Web. One application of crowdsourcing is to outsource certain tasks to the crowd that are difficult to implement in software. Another potential benefit of crowdsourcing is the on-demand allocation of a flexible workforce. Businesses may outsource tasks to the crowd based on temporary workload variations. A major challenge in crowdsourcing is to guarantee high-quality processing of tasks. We present a novel crowdsourcing marketplace that matches tasks to suitable workers based on auctions. The key to ensuring high quality lies in skilled members whose capabilities can be estimated correctly. We present a novel auction mechanism for skill evolution that helps to correctly estimate workers and to evolve skills that are needed. Evaluations show that this leads to improved crowdsourcing.

Crowdsourcing tasks to social networks in BPEL4People

Human-interactions are a substantial part of today’s business processes. In service-oriented systems this has led to specifications such as WS-HumanTask and BPEL4People which aim at standardizing the interaction protocol between software processes and humans. These specifications received considerable attention from major industry players due to their extensibility and interoperability. Recently, crowdsourcing has emerged as a new paradigm for leveraging a human workforce using Web technologies. We argue that crowdsourcing techniques and platforms could benefit from XML-based standards such as WS-HumanTask and BPEL4People as these specifications allow for extensibility and cross-platform operation. However, most efforts to model human interactions using BPEL4People focus on relatively static role models for selecting the right person to interact with. Thus, BPEL4People is not well suited for specifying and executing processes involving crowdsourcing of tasks to online communities. Here, we extend BPEL4People with non-functional properties that allow to cope with the inherent dynamics of crowdsourcing processes. Such properties include human capabilities and the level of skills. We discuss the formation of social networks that are particularly beneficial for processing extended BPEL4People tasks. Furthermore, we present novel approaches for the automated assignment of tasks to a social group. The feasibility of our approach is shown through a proof of concept implementation of various concepts as well as simulations and experiments to evaluate our ranking and selection approach.

Wind energy prediction and monitoring with neural computation

Wind energy has an important part to play as renewable energy resource in a sustainable world. For a reliable integration of wind energy high-dimensional wind time-series have to be analyzed. Fault analysis and prediction are an important aspect in this context. The objective of this work is to show how methods from neural computation can serve as forecasting and monitoring techniques, contributing to a successful integration of wind into sustainable and smart energy grids. We will employ support vector regression as prediction method for wind energy time-series. Furthermore, we will use dimension reduction techniques like self-organizing maps for monitoring of high-dimensional wind time-series. The methods are briefly introduced, related work is presented, and experimental case studies are exemplarily described. The experimental parts are based on real wind energy time-series data from the National Renewable Energy Laboratory (NREL) western wind resource data set.