[===================-----........ DAY 1 .........-----===================] [===================----- Start: 10:00 -----===================] [===================----- Session 1: 10:00-11:20 -----===================] Title: Repositories for Scientific Experiment Data -------------------------------------------------- Presenter: Rainer Stotzka Beamer: 1 This talk is about repositories for scientific experiment data. It starts with the definition of repositories/archives and shortly presents existing repository systems. Afterwards, requirements on a repository system for handling experiment data are described to contrast existing solutions with solutions needed for big data, e.g. in LSDMA. Title: KIT Data Manager: An Architecture for Building up Scientific Experiment Data Repositories ------------------------------------------------------------------------------ ----------------- Experiment Data Repositories Presenter: Thomas Jejkal Beamer: 1 This talk is about KIT Data Manager which provides services for building up respositories for scientific experiment data. In this presentation the architecture and basic services are described as well as their potential integration for setting up a repository system for a scientific community. Therefor, different scenarios are depicted depending on community requirements. [===================----- Coffee: 11:20-11:40 -----===================] [===================----- Session 2: 11:40-13:00 -----===================] Title: KIT Data Manager: An Example of Use ------------------------------------------ Presenter: Ajinkya Beamer: 2 This talk shows the integration of KIT Data Manager services for a real world application. The desciption of the requirements is followed by an overview of utilized KIT Data Manager services and community-specific customizations. Finally, the current state is demonstrated to the audience. Title: Overview over MongoDB ---------------------------- Presenter: Joerg (backup: Pari) Projectors: 2 (slides+shell) Abstract: The presentation will introduce the document-oriented NoSQL database MongoDB. Basic CRUD operations from different language drivers will be demonstrated. Advantages like the schema-less json-documents and disadvantages are discussed. Finally, applications of the DLCL Earth and environment are shown. Title: dCache, the Big Data cloud solution ------------------------------------------- Presenter: Patrick Fuhrmann (in case I don't find anyone else) Projector: just one For the past decade, the dCache technology has been continuously providing solutions for a wide range of scientific communities by closely collaborating with those communities and with middleware and infrastructure projects in Europe and the US (e.g. EGI, EMI, Open Science Grid). This presentation will give a brief overview of the available features in dCache to solve Big Data issues as well our current development in integrating dCache into the cloud ecosystem by providing the required interfaces and mechanisms terms of data access and user authentication. [===================----- Lunch: 13:00-14:00 -----===================] [===================----- Session 3: 14:00-15:30 -----===================] Title: LSDMA-AAI: using shared resources made easier ---------------------------------------------------- Presenter: Paul Millar Projectors: a second one might be useful. Science is increasingly done in collaborations, not just of people but involved aggregation of compute and storage resources from many institutes. As these resources are expensive, access must be controlled, but current methods provide barriers against such groups cooperating. In our demonstration we will simulate a small group of scientists, consisting of DESY and KIT employees. The group will cooperate in copying raw data from DESY to KIT. Once at KIT, it will be processed and the results copied back to DESY. All authentication will be with the users presenting only home institute credentials. Title: X.509 to SAML Credential Translation Service Prototype for LSDMA ----------------------------------------------------------------------- Presenter: Arsen Hayrapetyan This talk is to demonstrate a prototype implementation of an X.509 to SAML credential translation service. The service is a prerequisite for data management applications requiring SAML-based AAI for accessing resources protected by X509-based access control systems. The system expects user institutional credentials (username and password) and delivers short-lived X.509 certificates. An extension of the system is capable of generating grid proxy certificates to be used in virtual organisations. Title: Migrating legacy SP to LSDMA AAI --------------------------------------- Presenters: Dennis Klein (, Matthias Pausch) Projectors: 1 Status update on efforts on the migration of a web-based legacy SP to be integrated with the LSDMA AAI. The main problem addressed is the linking of legacy accounts to SAML identities in a self-service fashion from a user's perspective. [===================----- Coffee: 15:30-15:50 -----===================] [===================----- Session 4: 15:50-17:20 -----===================] Title: The Unity tool for Group Management ------------------------------------------ Presenter: Krzysztof Benedyczak In a large scale computing or data infrastructures, it is challenging (and indispensable) to support unified User's group management while keeping the federated identity intact. Thus, enable users to manage group and her attributes using the single federated identity. The talk introduces and demonstrates Unity - a federated users' group management system, builds on a concept of Authentication-As-a-Service (AaaS) is capable of not only managing groups, but offers federated identity management, multiple authentication protocols (e.g. SAML, PKI, LDAP, OpenId Connect, etc...), and fine grained group/attribute management. It also comes with sophisticated yet intuitive Web GUI for administrators as well as normal users. Title: The Http/WebDAV federation system ---------------------------------------- Presenter: Patrick Fuhrmann Projectors: just one With the growing trend of data intensive scientific experiments to distribute their data amount their partner laboratories, either for data redundancy reasons or to optimize data locations for fast distributed processing, simple mechanisms to access those distributed datasets are essential to enable efficient data analysis. This presentation will introduce an easy-to-use system, federating distributed data and making it available through the standard http/WebDAV protocol suite, allowing transfer-optimization either by server side plugins (e.g. GeoIP) or client side algorithms. The presentation will be followed by a short demonstration of one of the systems already serving data for WLCG from a variety of storage endpoints. [===================----- Orga Break: 17:20-18:00 -----==================] [===================----- Soccer: .pt -- .de -----==================] [===================-----........ DAY 2 .........-----===================] [===================----- Start: 9:00 -----===================] [===================----- Session 5: 9:00-10:20 -----===================] Title: UNICORE -------------- Presenter: Bernd Schuller UNICORE is a federation software suite, providing secure and easy-to-use access to compute clusters, HPC systems and data resources. UNICORE includes submission and management of simple jobs and complex workflows as well as a complete set of data management functions including metadata management and rule-oriented processing. This presentation will provide a walk-through through UNICORE and its main features. Title: STXXL 1.4.0 and Beyond: External Memory Algorithms --------------------------------------------------------- Presenter: Timo Bingmann, KIT In this talk we give an introductory tutorial into STXXL, a library of basic external memory algorithms with STL-compatible interfaces. With this library, many C++ applications can be extended to larger datasets easily. After skimming hardware properties and the external memory model, we show how most STL containers are mapped to external memory by STXXL. Then, we focus on the most important external memory algorithm implemented in STXXL: sorting. With these basics laid out, we discuss applications which we and others developed with STXXL in the past, and give an outlook into future extensions, applications and research. [===================----- Coffee: 10:20-10:40 -----===================] [===================----- Session 5: 10:40-12:00 -----===================] Title: UFTP ----------- Presenter: Bernd Schuller UFTP is a file transfer tool similar to GridFTP, but without the security flaws (e.g. requirement for open port ranges in the firewall). It is on the one hand integrated with UNICORE, but on the other hand can also be used standalone. This presentation focuses on the latter case, showing the standalone client and includes a discussion of deployment options. Title: Power-performance tracing framework ------------------------------------------ Presenter: Manuel Dolz Projectors: 1 Abstract: In this demo we present an evolved framework to trace and analyze the performance and energy consumption of parallel scientific applications that works in combination with Vampir and VampirTrace. The framework includes i) a flexible and extensible design that enables easy integration of different types of power measurement devices; and ii) modules that record resource utilization values, such as disk and network throughput.