- Sponsor:
- sigops
Relational databases have been for a long time the keystone of information systems' dependability. Database management systems (DBMS) offer a uniform approach to data integrity, durability, and availability, using tried and tested techniques based on a set of unanimously accepted and well understood assumptions.
However, the expectations on the DBMS itself are changing and there is an unprecedented call for a complete rewrite: The general purpose relational DBMS architecture dating back from the 70's is being convincingly challenged by a number of specialized systems for text indexing, data warehousing, stream processing, and array storage. Most strikingly, the winds of change are felt even within the common enterprise data center, where the widespread need for practical scale-out and near zero downtime translates to an increased appetite for cheap and efficient consistent replication and for shared-nothing clusters built on commodity hardware and software.
Novel peer-to-peer applications, grid computing, and the emergence of cloud computing further push the envelope for radically different data management solutions. A major trend in all such emerging proposals, with a profound impact in dependability, is that large scale distribution is a core assumption in their design. This translates in very large number of nodes, wide area geographical distribution, diverse administrative domains, and pervasive heterogeneity.
It is thus desirable to reevaluate time tested assumptions that underly the dependability mechanisms in database management systems and explore different performance and functionality tradeoffs. And explore the consequences on complex information systems built on them.
Proceeding Downloads
Adaptive replication control based on consensus
This paper presents a meta protocol that allows the dynamic replacement of replication control protocols in replicated databases. The meta protocol is motivated by the diversity of concurrency control and replication control protocols that implement ...
dsmDB: a distributed shared memory approach for building replicated database systems
Current trends in main memory capacity and cost indicate that in a few years most performance-critical applications will have all (or most of) their data stored in the main memory of the nodes of a small-size cluster. A few recent research papers have ...
Versioned transactional shared memory for the FénixEDU web application
The FénixEDU system uses a novel infrastructure for web applications based on the Versioned Software Transactional Memory (VSTM) abstraction. The FénixEDU system has been deployed and is currently in operation in different facilities, including the ...
ExoSnap: a modular approach to semantic synchronization and snapshots
Current approaches to application-specific synchronization suffer from a limitation that precludes the use of generic "commodity" servers because they require to run the type-specific synchronization code at the servers. This is a problem for "cloud ...
Challenges in dependable internet-scale stream processing
Today we lack an infrastructure for globally processing stream data from sensor networks and making this data available to millions of users in real-time. To build such a system, we need to address a set of challenges and, in particular, rethink what ...
- Proceedings of the 2nd workshop on Dependable distributed data management