Hintergrundbild © Stuttgart-Marketing GmbH
Operating and maintaining a large industrial infrastructure such as the one at CERN creates a lot of challenging data problems.
As both data volumes and rates increase, non-functional requirements such as performance, availability and maintainability are getting more important than ever. Beside commercial off-the-shelf solutions CERN has developed a modular Open Source Java framework called C2MON for realising highly available, large industrial monitoring and control solutions. It has been initially developed for CERN’s demanding infrastructure monitoring needs and is based on more than 10 years of experience with the Technical Infrastructure Monitoring (TIM) systems at CERN. Combining maintainability, high-availability and a harmonized data structure within a portable architecture is the focus of this work. Making use of standard Java libraries for in-memory data management, clustering and data persistence, the platform becomes interesting for many Big Data scenarios.
Attendees will get an insight of the CERN specific solutions to handle data acquisition, analysis and storage of the various technical equipment around Large Hadron Collider (LHC). The talk will cover the ongoing group activity to modernise the technology stack with regard to the constant increase of data volume, variety and velocity. More specific examples will be elaborated in the second half of the presentation with a focus on existing C2MON use cases.
About: Matthias Braeger is Software Engineer (staff) at CERN, the European Organization for Nuclear Research. He is responsible for the Technical Infrastructure Monitoring (TIM) system which is a 24/7 service used by many different user groups to supervise and control a variety of infrastructure spread across the CERN site. He also heads the CERN Control and Monitoring Platform C2MON (http://cern.ch/c2mon) which allows the rapid development of highly reliable, scalable and flexible monitoring solutions. Before joining CERN, Matthias worked for Logica and helped developing the Spacecraft Control & Operation System for ESA. Matthias is author of several scientific papers and a featured speaker at industry conferences and training seminars.
Abstract: Answer Set Programming (ASP) is a well established approach to declarative problem solving, combining a rich yet simple modeling language with high-performance solving capacities. In this talk we present asprin, a general, flexible and extensible framework for preferences in ASP. asprin is general and captures many of the existing approaches to preferences. It is flexible, because it allows for the combination of different types of preferences. It is also extensible, allowing for an easy implementation of new approaches to preferences. Since it is straightforward to capture propositional theories and constraint satisfaction problems in ASP, the framework is also relevant to optimization in Satisfiability Testing and Constraint Processing.
About: Javier Romero is a researcher at the University of Potsdam, Germany. He received a M. Sc. in Computer Science from the University of Corunna in 2006, and in Philosophy from the University of Santiago de Compostela in 2008, both in Spain. His research interests are knowledge representation and logic programming, and his research focuses in extensions of Answer set programming (ASP) for declarative heuristics and preference reasoning. He is a member of the open source project potassco.org of ASP tools developed at Potsdam, and the main developer of asprin, a system for preferences in ASP.
Scalable cloud data management covers many aspects. Over the last decade, Big Data companies and the research community alike have made great progress in many of these: We know how to scale out data stores, how to execute MapReduce queries, and we have learnt a lot about how to tune them. However, there are also aspects that practitioners are actively struggling with: When it comes to continuously deploying applications against schema-flexible NoSQL data stores, there is very little shared experience in the community on what works and what doesn’t.
In this keynote, we address the challenge of schema evolution and NoSQL data migration in this context. We start with the state-of-the-art in schema management for NoSQL databases, and continue with outlining the subtasks of a controlled and systematic schema management. We then focus on schema evolution and scalable NoSQL data migration. We distinguish two core strategies, namely to migrate all legacy data eagerly, in one go, or to lazily migrate individual persisted objects when they are loaded by the application. We also introduce variations that blend eager and the lazy migration. To scale to large datasets, we explore different optimizations to the data migration process, ranging from a rule-based composition of data migration operations up to a predictive data migration, based on forecasts which datasets are likely to be accessed in the near future. We close the keynote with an overview on open research questions and our plans for future work.