JPPF makes it easy to parallelize computationally intensive tasks and execute them on a Grid
JPPF 6.1 alpha 2 new features highlights
Asynchronous communication between node and server
- nodes can now process any number of jobs concurrently
- this addresses the issue where some threads of a node could be idle, even when work was available
- overall performance gain due to I/O being performed in parallel with job processing
- configurable limit on the number of jobs a node can process concurrently (unlimited by default)
Pluggable node throttling mechanism
- the main purpose of this mechanism is to avoid resource (e.g. heap, cpu) exhaustion in the node, due to too many concurrent jobs. When a condition is reached, the node refuses new jobs. When the condition no longer applies, the node accepts jobs again.
- it can also be used to limit the workload based on any condition, for instance based on time windows or the status of external services
- A built-in implementation provides throttling based on heap usage
New job SLA attributes
- maximum driver depth: in a multi server topology, an upper bound for how many drivers a job can be transfered to before being executed on a node (server-side SLA)
- maximum dispatch size: the maximum number of tasks in a job that can be sent at once to a node (driver-side SLA) or to a driver (client-side SLA). This overrides the dipsatch size computed by the load-balancer
- allow multiple dispatches to the same node (driver-side SLA) or driver (client-side SLA): a flag to specifiy whether a job can be dispatched to the same node or driver multiple times at any given moment
Offline documentation (PDF)
- the source documents (.odt) were regrouped into a single document, in order to fix the broken cross-document links in the produced PDF
- the font and background for code snippets were changed to improve readability
All implemented issues
Features
- JPPF-564 Asynchronous communication betwen node and driver
- JPPF-569 New job SLA attributes
- JPPF-570 Accessing the job from a task
- JPPF-573 Pluggable mechanism to warn the driver when a node can't accept any more job
Enhancements
Tasks
Links
Previous JPPF 6.1 release announcements: JPPF 6.1 alpha
JPPF Links: Web Site - Downloads - Documentation - User forums - Issue tracker