News

• Final program available.

Context and Goal

Database replication is widely used to improve both the performance and resilience of database management systems. Although most commercially available solutions and the large majority of deployments use asynchronous updates in a shared nothing architecture, there is an increasing demand for additional guarantees, configuration flexibility, and manageability, which are currently provided only, if at all, by costly high-end solutions.

An example is the desire to upgrade current fail-over clusters to an active-active configuration, thus leveraging the additional computational resources for additional performance. It should also be possible to configure eager update replication over a MAN or WAN for disaster recovery, without resorting to volume replication, thus improving manageability and performance. The combination of both local clusters and wide-area systems in grid-style large scale systems poses new challenges in performance and manageability. The drive for self-manageable systems also discourages the need for human intervention in recovery and conflict solving, thus favoring eager update approaches. Finally, typical loads of current middleware and applications might allow novel trade-offs between resilience and performance that are not viable with OLTP loads.

This demand has been addressed by a flock of proposals for eager update replication that do not use the traditional 2PC-based approach. These range from middleware based approaches, which often also provide load balancing of read-only queries, to integrated solutions as demonstrated in the Postgres-R prototype. Finally, there is a flourishing academic research trend in group communication based replication, which offloads many complexity and performance issues to group communication toolkits and aims at scalable eager update everywhere replication.

The goal of the workshop is to bring together researchers and practitioners from the database and fault-tolerant distributed systems communities to discuss the current state of the art, pending challenges and trends, and novel solutions in the design, implementation and deployment of database replication.

Program

Location: Brattøra

Session 1 - 9h15-10h30

• Welcome/About the program
• Revisiting the database state machine approach
  V. Zuikeviciute, F. Pedone (Università della Svizzera Italiana)
• Exactly once interaction in a multi-tier architecture
  B. Kemme, R. Jimenez, M. Patiño, J. Salas (McGill University, Universidad Politécnica de Madrid)

Coffee break

Session 2 - 11h00-12h30

• Oracle® Streams for near real time asynchronous replication
  N. Arora (Oracle)
• Replication tools in the MADIS middleware
  L. Irún-Britz, J. Armendáriz, H. Decker, J. Mendívil, F. Muñoz-Escoí (Universidad Politécnica de Valencia, Universidad Pública de Navarra)
• Replication in node partitioned data warehouses
  P. Furtado (University of Coimbra)

Lunch break

Session 3 - 14h00-15h30

• Large-scale experimentation with preventive replication in a database cluster
  P. Valduriez, E. Pacitti, C. Coulon (University of Nantes)
• Fine-grained refresh strategies for managing replication in database clusters
  S. Gançarski, H. Naacke, C. Le Pape (Laboratoire d´Informatique de Paris 6)
• Self-manageable replicated servers
  C. Taton, S. Bouchenak, F. Boyer, N. De Palma, D. Hagimont, A. Mos (INRIA, University of Grenoble I)

Coffee break

Session 4 - 16h00-17h00

• Panel
  R. Jimenez (Universidad Politécnica de Madrid, moderator), N. Arora (Oracle), E. Cecchet (EMIC Networks), P. Valduriez (University of Nantes), J. Pereira (University of Minho)

Registration

Registration for the workshop is available through the main VLDB registration page.

Committees

• Program Chair:
  • R. Oliveira (Universidade do Minho, Portugal)

• Organizing Chair:
  • J. Pereira (Universidade do Minho, Portugal)
• Program Committee:
  • Emmanuel Cecchet (EMIC Networks & ObjectWeb Consortium, Finland)
  • Jonathan Goldstein (Microsoft Research, USA)
  • Bettina Kemme (McGill University, Canada)
  • Thomas Lane (RedHat Inc., USA)
  • Henrique Madeira (Universidade de Coimbra, Portugal)
  • Marta Patiño-Martínez (Universidad Politécnica de Madrid, Spain)
  • Fernando Pedone (Università della Svizzera Italiana, Switzerland)
  • Luís Rodrigues (Universidade de Lisboa, Portugal)
  • Jonathan Stanton (George Washington University, USA)
  • Jean-Bernard Stefani (INRIA, France)
  • Lars Thalmann (MySQL AB)
  • Jan Wieck (Affilias Inc., USA)

Call for Papers

• Paper submission is now closed. The Call for Papers is still available on-line, as plain text or PDF.

Additional Information

The workshop is organized in assotiation with the VLDB conference by the GORDA project (FP6-IST2-004758). For more informations contact the organization at datarep@lsd.di.uminho.pt.