Red Hat has launched a real-time version of Linux, aimed at applications needing predictable responses, from process control to financial market traders.

Red Hat Enterprise MRG (Messaging, Realtime, Grid) includes features such as high-speed inter-application messaging based on the Advanced Message Queuing Protocol (AMQP), whose performance RH reckoned it had improved 100-fold. At the UK launch, RH VP Scott Crenshaw could not confirm how this was achieved, representatives said that details will be available on the company's blog.

MRG also allows you to "steal" unused desktop CPU cycles, to manage distributed workloads, to schedule tasks across both local and remote grids, and use cloud capacity from Amazon EC2. The distributed computing capabilities emanate from RH's collaboration with the University of Wisconsin and its high-throughput computing project, Condor. The code for this portion of the system is open source under an OSI-approved licence.

While MRG can be run on Java, Solaris and .Net platforms, the company said that best performance will be obtained when running on Red Hat Enterprise Linux (RHEL).

According to Crenshaw, the first customer for the system was the US Navy. "They approached us a few years ago for a system that could run whole ships, weapons control, the lot," he said.

"As a working group member of AMQP, Cisco has been collaborating with Red Hat for over 18 months in low-latency optimisation of AMQP and MRG Messaging open middleware across Infiniband compute fabrics," said Cisco marketing director Bill Erdman.

According to RH, the messaging and grid capabilities can be deployed in multiple environments and are optimised for use with RH's real-time Linux.

Crenshaw said that pricing had yet to be announced.