August 7 – 12, 2011, Dagstuhl Seminar 11321
Information Management in the Cloud
Anastasia Ailamaki (EPFL – Lausanne, CH)
Michael J. Carey (University of California – Irvine, US)
Donald Kossmann (ETH Zürich, CH)
Steve Loughran (HP Lab – Bristol, GB)
Volker Markl (TU Berlin, DE)
Raghu Ramakrishnan (Microsoft Research – Redmond, US)
1 / 2 >
For support, please contact
Art exhibition opens on Tuesday August 9
All participants are invited to attend after dinner on 7:30 pm.
More information here .
Cloud computing is emerging as a new paradigm for highly scalable, fault-tolerant, and adaptable computing on large clusters of off-the-shelf computers. Cloud architectures strive to massively parallelize complex processing tasks through a computational model motivated by functional programming. They provide highly available storage and compute capacity through distribution and redundancy. Most importantly, Cloud architectures adapt to changing requirements by dynamically provisioning new (virtualized) compute or storage nodes. Economies of scale enable cloud providers to provide compute and storage powers to a multitude of users. On the infrastructure side, such a model has been pioneered by Amazon with EC2, whereas software as a service on cloud infrastructures with multi-tenancy has been pioneered by Salesforce.com.
The Dagstuhl seminar on Information Management in the Cloud brought together a diverse set of researchers and practitioners with a broad range of expertise. The purpose of this seminar was to consider and to discuss causes, opportunities, and solutions for technologies, and architectures that enable cloud information management. The scope ranged from web-scale log file analysis using cluster computing techniques to dynamic provisioning of resources in data centers, covering topics from the areas of analytical and transactional processing, parallelization of large scale data and compute intensive operations as well as implementation techniques for fault tolerance.
The seminar consisted of keynotes, participant presentations, demos and working groups. The first two seminar days consisted of a keynote by Helmut Krcmar on ``Business Aspects of Cloud Computing'' as well as 33 short presentations on various aspects of cloud computing. On the evening of the second day, the participants formed working groups on economic aspects, programming models, benchmarking. The third day of the seminar consisted of two keynotes, by Dirk Riehle on ``Open Source and Cloud Computing" and by Donald Kossmann on ``Benchmarking''. After these keynotes, working groups discussed their respective topics. In the evening, an industrial panel with Miron Livny, Steve Loughran, Sergey Melnik, Russell Sears, and Dean Jacobs discussed research challenges in Cloud Computing from an industrial point of view. On the fourth day, a keynote by Miron Livny discussed Cloud Computing from a distributed systems and high-performance computing point of way. After the keynote, a demo session presented the following systems:
- HyPer: A Cloud-scale Main Memory Database System (Team from TUM)
- Asterix and Hyrax (Team from UCI)
- Stratosphere (Team from TU Berlin, HU Berlin and HPI)
- Myriad Parallel Data Generator (Team from TU Berlin)
After these demos, working groups continued during the day and presented their results in the evening. The last day of the seminar, participants continued in working groups and discussed further collaborations with respect to papers and project proposals. During this day, several abstracts for papers have been prepared, and discussions about several joint research project proposals have started.
The organizers hope that the seminar has helped to organize the research space in cloud computing and identified new research challenges. We look forward towards research collaborations and papers that were bootstrapped during this intensive week.
- Databases / Information Retrieval
- Data Structures / Distributed Computing / Scheduling
- Autonomic Computing / IT-management
- Cloud computing
- Information management
- Software as a service
- Platform as a service
- Grid computing
- Cluster computing
- Query processing