Warning: Parameter 1 to Language::getMagic() expected to be a reference, value given in /usr/share/mediawiki/includes/StubObject.php on line 58
Workshop:Data Services – GRK-Wiki

Workshop:Data Services

Aus GRK-Wiki

Wechseln zu: Navigation, Suche

Inhaltsverzeichnis

3. Metrik-Workshop: Data and Service Management in Self-Organizing Systems

Das Graduiertenkolleg Metrik lädt am 11. und 12. Oktober zum Workshop Data and Service Management in Self-Organizing Systems ein.

Der Workshop beleuchtet die Anwendung Service-orientierter Architekturen und von Publish-Subscribe-Systemen zur Selbstorganisation (großer) verteilter Systeme. Die Gastredner Gustavo Alonso und Arno Jacobsen stellen Grundlagen und praktische Anwendungen dieser Paradigmen vor. In weiteren Vorträgen des Graduiertenkollegs METRIK, der HPI Research School on "Service-Oriented Systems Engineering" und der University of Waikato diskutieren wir daran anschließende Fragestellungen und Lösungsansätze.

Koordinaten

  • Datum:
    11. und 12. Oktober 2007
  • Ort:
    Humboldt-Universität zu Berlin
    Institut für Informatik
    Humboldt-Kabinett
    Rudower Chaussee 25

Programm

Donnerstag, 11. Oktober 2007
10:00-17:00 Humboldt-Kabinett  
 
10:00   Gustavo Alonso : Software Adaptation for large scale SOA
11:30 Stefan Brüning : Dependable, Service-Based Infrastructures for Disaster Management
12:00 Mittagspause
13:00 Arno Jacobsen : The PADRES Enterprise Service Bus
14:30 Artin Avanes : Transactional Workflows in Self-Organizing Systems
15:00 Pause
15:30 Annika Hinze : Service Collaboration in Location-aware Applications
16:30 Diskussion
 
ab 20:00 Abendessen
Freitag, 12. Oktober 2007
10:00-17:00 Humboldt-Kabinett  
 
10:00   Gustavo Alonso : The Service Revolution - software engineering without the burden of programming languages
11:30 Timo Mika Gläßer : Query Processing for Wireless Sensor Networks
12:00 Mittagspause
13:00 Arno Jacobsen : Content-based Routing in General Overlay Topologies
14:30 Bastian Quilitz : Semantic Integration of Geographic Data
15:00 Pause
15:30 Andreas Polze : Operating System Support for Service Computing
16:00 Matthias Uflacker : Resource-oriented Knowledge Networks
16:30 Diskussion

Vorträge

Gustavo Alonso: Software Adaptation for large scale SOA

Building systems of systems poses many different and difficult challenges. In particular, it does away with the notion of static software that does not need to change between different deployment cycles. The problem becomes even more complex since realistic requirements dictate that changes to the software to adapt it to different conditions must take place at run-time and without stopping the application. In this talk I will focus on the problem of run-time software adaptation and describe two of our efforts in the area: PROSE and R-OSGi. PROSE is a dynamic Aspect Oriented Programing tool that allows arbitrary changes to a running Java program. R-OSGi is an extension to the OSGi specification to support the distribution at the level of software modules and allow developers much more control over the dynamic evolution of their software.

Gustavo Alonso: The Service Revolution - software engineering without the burden of programming languages

Service Oriented Architectures (SOA) encompass a wide range of technologies. In this talk I will focus on the impact of SOA on application development and how it shifts the emphasis of software engineering from programming to integration. As part of this process, a wealth of new opportunities arise to tackle a variety of open problems that software engineering has not been able to address in the last decades: formal treatment of non-functional requirements, continuous development, dynamic adaptation, and new forms of data storage and delivery. In the talk I will outline how SOA provides us with the necessary abstractions to successfully address these key problems. I will also discuss in detail what are the key contributions of the concepts around SOA and underline their importance for future developments in the area.

Arno Jacobsen: The PADRES Enterprise Service Bus

The PADRES research project develops an Enterprise Service Bus (ESB) based on the idea of decentralized service composition resulting in agile software system architectures. The PADRES ESB can serve as basis for building service-oriented and event-driven enterprise architectures. PADRES offers content-based message routing, in-network filtering, event correlation, unification of future and historic data access, load-balanced messaging, and failure resilience. These capabilities enable PADRES to transparently manage infrastructure and application events, resulting in a powerful solution for business activity monitoring, execution and management. The architecture of PADRES is based on a distributed content-based publish/subscribe system, where content-based message routing brokers are connected in an application-oriented overlay topology. The basic functionality of PADRES is the routing of data messages (publications) from data source clients to interested data sink clients. PADRES uses content-based routing to effectively decouple publishers and subscribers. In this talk, I provide an overview of event processing based on publish/subscribe comparing it to database processing, motivate various application scenarios well suited for publish/subscribe-style processing, present the PADRES architecture and system, and discuss the decentralized execution of business processes with PADRES.

Arno Jacobsen: Content-based Routing in General Overlay Topologies

Content-based routing is well suited as messaging substrate to support distributed applications. To simplify the design, most content-based routing approaches assume a tree-based overlay network, or use protocols where routing brokers perform controlled flooding which is expensive and entails duplicate messages. We developed techniques to extend content-based routing to accommodate general overlay topologies. Among other benefits, our approach allows the protocol to adapt to dynamic network conditions, such as congested or failed routes, by choosing among alternative routing paths. Our design maintains the original simple publish and subscribe protocol interface to clients and is easy to apply to existing approaches. In this talk, I provide an overview of content-based routing, covering atomic subscription and composite subscription routing, and extend the routing algorithm to accommodate general overlay topologies. In particular, I focus on how to exploit the flexibility gained from supporting alternative routing paths for adaptive composite subscription routing. I illustrate a dynamic composite event detection algorithm that optimizes the cost of evaluating composite subscriptions with respect to network traffic and message delay. The effectiveness of the approach is demonstrated by reviewing experiments conducted on a cluster of machines and on PlanetLab.

Persönliche Werkzeuge
Sprache