FastMPJ


High Performance Java Message Passing Library

Fast Java Messaging (FastMPJ) is the fastest Java messaging library, an implementation of Message-Passing in Java for high performance support on shared memory and high speed Remote Direct Memory Access (RDMA) networks such as InfiniBand.

 

FastMPJ is the leading solution in the industry, being the only fully portable, as Java, but also having the best performance, similar to C MPI. Development and maintenance of FastMPJ applications is highly productive, while providing high performance intra-node communication and efficient RDMA transfers over Infiniband and RoCE.

 

FastMPJ has been carefully designed to obtain scalability up to thousands of cores. Thus, being an ideal solution for multicore servers, cluster and cloud computing.

 

Being especially suitable for High Performance computing, FastMPJ is also an excellent choice for financial sectors as also to Web 2.0 applications.


Improve Performance of Financial Market Data

The transparent and high performance nature of Fast Java Messaging provides a measurable increase in application performance to those with heavy traffic volumes or low latency messaging requirements. Thus, ideal candidates for FastMPJ include:

 

  • Market data consumption and distribution applications
  • High frequency trading and real time risk management
  • Caching/data distribution application


Increase Throughput of Cloud and Web 2.0 Applications

As data volumes increase in a variety of enterprises, market segments and applications, FastMPJ becomes a popular choice for improvement of Web 2.0 applications performance. Ideal candidates include:

 

  • Distributed memory caching systems
  • Multi-sensor data acquisition processing
  • Online content delivery and caching


Key Features

Features

  • Fully portable, as Java
  • Highly productive development and maintenance
  • Ideal for multicore servers, cluster and cloud computing

 
Performance
  • Java High Performance Computing (HPC)
  • Efficient RDMA transfers over IB and RoCE
  • Very Low Latency
  • High Throughput
  • Scalability up to thousands of cores

 
Connectivity
  • Shared memory
  • High speed networks


fmpj_shm_performancefmpj_network_performance