Document Actions
CASTOR
Causality Tracking for Optimistic Replication in Dynamic Distributed Systems
The goal with project Castor, funded by the Portuguese Foundation for Science and Technology (FCT), was to improve the technology supporting Cloud systems subjected to highly dynamic environments, not only in terms of the number of clients, but also in terms of the number of computer nodes.
In particular, the team wanted to improve the causality tracking system, which is key when designing relaxed coherence systems, and to improve the causality tracking and reconciliation management features inspired by Amazon’s Dynamo.
As a result, the team developed new causality tracking techniques, Dotted Version Vectors and DVV Sets, as well as unified and deterministic data processing via Conflict-Free Replicated Data-Types (CRDT), in a collaboration with the French project Concordant.
Currently, these two techniques are incorporated in a commercial database, Basho’s Riak, and the CRDT technology is available in different languages and systems, namely in Typesafe’s Akka Data Replication, in Soundcloud’s Roshi, and in Facebook’s Apollo.
Forgot your password?
