A fundamental component of the MRG package is the real-time Red Hat kernel, which is valuable for applications such as stock trading or missile systems, said Bryan Che, Red Hat's MRG product manager. Rather than referring to speed, real-time refers to low latency or a guaranteed response time, he said. Real time still poses tradeoffs in performance levels, so it's not the best option for every application, he said.
The messaging component, meanwhile, is built on the Advanced Messaging Queuing Protocol (AMQP) protocol, which can handle multiple kinds of messaging functions that otherwise require separate products, depending on whether speed, reliability or another function is most important, Che said. AMQP, however, can handle any messaging task.
In addition, Cisco has developed complementary networking protocols that enable certain types of AMQP messages to be prioritized over other messages, which is helpful with high-volume applications with sudden peak loads, such as stock market trading, he said.
The new messaging component is hardware-neutral and highly optimized for Linux, he added. It will also be supported as part of Red Hat's JBoss Enterprise SOA platform.
Grid, the final component of MRG, will enable companies to expand their computing capacity on demand, in remote locations or on a cloud such as Amazon's Elastic Compute Cloud, or EC2. Now, with the help of Red Hat's virtualization management infrastructure, just released in beta format, companies have the tools to build their own corporate on-demand "clouds," that IT can manage and apportion directly.
Further, grids can help with excess capacity or enable companies to minimize their data center resources and relegate the rest to on-demand cloud computing, Che said. Availability of the grid offering, based on the open source Condor grid management project, is expected by the end of the year.
The three capabilities of MRG are useful individually but are especially powerful in combination, Che said.