What's new in JPPF 2.1
PerformanceThe development of this release has seen a lot of emphasis on both memory usage and performance improvements.
The JPPF server now executes multithreaded network I/O and reaches new levels of scalability.
JPPF applications can now enjoy faster execution of larger jobs.
LocalizationRussian localization is now available for the graphical administration console.
Load balancingCustom load-balancers can now use information about the nodes' environment and configuration, along with metadata about the jobs.
This allows for sophisticated load-balancing implementations that can adapt to the capabilities of each node and the computational characteristics of each job.
A sophisticated and fully documented sample illustrates these new capabilities and will get you started in no time.
ConfigurationThe managment port auto-incrementation enables servers and nodes to automatically find available port numbers, making JPPF configuration even easier and removing one the main configuration hurdles.
It is now possible to specify the number of concurrent threads performing I/O in the server.
New SamplesThree new samples complement our offering in the JPPF samples pack:
- Simulation of large portfolio updates
- JPPF node health monitor in the system tray
- An example of a sophisticated load-balancer implementation
New Feature Requests
- 2966065 - Adapt the load-balancing to the weight of each job
- 2955505 - Give load balancers access to the node's configuration
- 2182052 - Simple local node monitoring in system tray
- 2972979 - Un-connected nodes shouldn't report a valid number of thread
- 2969881 - Undefined processing.threads causes execution policy to fail
- 2969126 - ClientDataProvider feature not implemented in J2EE connector
- 2967151 - Improper exception handling in JPPFBroadcaster
- 2962404 - Changing load-balancer settings is done at the wrong time
- 2955491 - Connection pool not working whith server discovery disabled
- 2953562 - Node does not report results of tasks with timeout
- 2933677 - Server thread stuck when serialization fails in node
- 2914622 - Local IP addresses should include more than 127.0.0.1
- 2907258 - JMX initialization failure causes node to stop working
- 2907246 - Remote debugging is only possible on localhost
(Note: Opinions expressed in this article and its replies are the opinions of their respective authors and not those of DZone, Inc.)