Cluster and Grid computing made easy (Jackal)
Internally funded project
Acronym: Jackal
Start date : 01.01.2006
End date : 31.12.2010
Project details
Scientific Abstract
Jackal is a project to create a distributed-shared-memory system for Java. This means that you can write a multi-threaded program (that you could run using normal JVMs on single machines as well) and deploy it on a cluster connected by a network. Jackal also sports a nice check-pointer so it can periodically write the program state to disk for fault tolerance.
To make things more interesting, you can write the program also using OpenMP annotations which allows re-parallelization. Combined with checkpointing this allows a program to be restarted on a different number of machines as that it was started with.
The Jackal-DSM is not only suited for traditional clusters where each node containes only a single CPU and core, but also for newer style clusters where each node in the cluster contains a multi-core CPU. Additionally, Jackal also has extensions and tools for Grid computing.