Traditionally, query optimizers rely on cost models to choose the best execution plan from several candidates, making precise cost estimates critical for efficient query execution. In recent years, cost models based on machine learning have been proposed to overcome the weaknesses of traditional cost models. While these models have been shown to provide better prediction accuracy, only limited efforts have been made to investigate how well Learned Cost Models (LCMs) actually perform in query optimization and how they affect overall query performance. In this paper, we address this by a systematic study evaluating LCMs on three of the core query optimization tasks: join ordering, access path selection, and physical operator selection . In our study, we compare seven state-of-the-art LCMs to a traditional cost model and, surprisingly, find that the traditional model often still outperforms LCMs in these tasks. We conclude by highlighting major takeaways and recommendations to guide future research toward making LCMs more effective for query optimization.

In this work, we present Costream, a novel learned cost model for Distributed Stream Processing Systems that provides accurate predictions of the execution costs of a streaming query in an edge-cloud environment. The cost model can be used to find an initial placement of operators across heterogeneous hardware, which is particularly important in these environments. In our evaluation, we demonstrate that Costream can produce highly accurate cost estimates for the initial operator placement and even generalize to unseen placements, queries, and hardware. When using Costream to optimize the placements of streaming operators, a median speedup of around 21 × can be achieved compared to baselines.

Parallel database systems are major platforms for supporting analytical queries over large data sets. However, in order to offer SQL-like services for data analytics in the cloud, providers such as Amazon and Google do often build their own systems (e.g., BigTable). One reason is that existing database systems do not fulfill important requirements such as elasticity and fine-grained fault-tolerance. In this poster, we present XDB [2, 3], a parallel database system which implements two novel concepts: (1) a partitioning scheme that supports elasticity with regard to data and queries, and (2) a fine-grained fault-tolerance scheme for short- and long-running queries.

Scientists are dealing with a larger and more complex IT infrastructure in their daily work. With the advent of Grid and Cloud technologies this trend continues. Scientists will not only deal with local managed resources, but with resources provided by an arbitrary Cloud service provider. In order to support the scientist in his daily work a scientific workbench is required to interact with IT resources and to perform scientific analysis on them. The g-Eclipse framework is such a platform for Grid/Cloud computing based on the Eclipse ecosystem. Based on the g-Eclipse/Eclipse architecture new scientific applications/tools can be integrated in this workbench. The g-Eclipse workbench is an ideal starting point for building such a workbench for astro particle physics. The usage of remote Cloud resources requires the development of an interoperable communication and software provisioning system too. Using OSGi as a interoperable, provider-independent platform for Cloud computing, the management of the Cloud resources will be simplified significantly. The delivery of software on the scientists computer and on his Cloud resources with a ScienceStore/ScienStore component will be as easy as the deployment of little applications on end customers smart phones.

The vision of cross-domain file systems is becoming a reality; however, issues like ensuring quality of service for bandwidth-intense resource access, especially over mobile network, may still arise. In this paper, we propose a cross-domain shared file system, which enables a real-time experience for delivery of virtualized file systems utilizing a grid approach over NGN. The system allows users to share file systems in real-time, and provides functions to delivery update notification, to ensure access control settings, and to offer real-time streaming access to the virtual resources. We describe the system design, and a file sharing scenario to show the details of the prototyped solution. Finally, we give an outlook on the future improvements that could be made to the system.

The convergence between grid environments and QoS-enabled networks, in particular in the context of next generation networks (NGN), gained research interest in the direction of combining NGN capabilities and IT assets for the creation of future services. Following the ETSI NGN grid standardization activities, this work proposes an OGSA standard compliant NGN grid subsystem architecture to act as a mediator between grid middleware and NGN service stratum. The novel step of this subsystem is the virtualization of NGN endpoints via grid services. We show that the enhancement of the NGN service delivery platform by the integration of grid technologies enables a new way of introducing distributed multimedia services.