Red Hat Ready for Top End of Enterprise Servers
Red Hat claims that Red Hat Enterprise MRG (pronounced merge) is a revolutionary distributed computing platform that will provide record-break grid performance through revolutionary new techniques in scheduling, messaging, real-time low-transaction latency and distributed computing.
Specifically, MRG will do this by enabling enterprises to schedule large computing tasks across local grids, remote grids, "cloud" capacity from Amazon EC2 and idle desktop workstations. So, for example, according to Bryan Che, MRG's product manager, MRG will be able to make use not just of RHEL servers on the network or in the cloud but on idle Windows workstations. With support for multiple versions of Linux, Solaris and Windows, if you can find an idle cycle on a computer, you'll find a computer that MRG can use for its work.
For messaging: Red Hat Enterprise MRG provides reliable messaging technologies. Scott Crenshaw, Red Hat's vice president of enterprise Linux, claims that it can deliver message performance up to 100 times faster than such propritary approaches as Tibco's TIBCO Enterprise Message Service. This capability comes from messaging technology developed with the open standard AMQP (Advanced Message Queuing Performance) Working Group. Che specifically claimed that MRG's AMQP will be capable of delivering 500,000 messages per second.
AMQP is an attempt to create an open standard for high-performance, reliable messaging. It is supported by hardware network vendors, Cisco; operating system companies, Red Hat and Novell; and middleware companies, Iona. AMQP, however, started out as a customer-driven initiative by JPMorgan. JPMogan and other high-end financial companies such as Credit Suisse and Goldman Sachs wanted an open-standard for messaging both to foster innovation in message-driven financial middleware and applications and to drive down software prices.
Red Hat Enterprise MRG real-time capabilities also incorporate what Crenshaw calls low-transaction latency. This enables applications to run with optimized and deterministic latency. In other words, besides helping provide HPC (high-performance computing), it also ensures that transactions run predictably under all workloads. Crenshaw called this HTC (high-throughput computing). With Enterprise MRG real-time capabilities, CIOs can match compute capacity to business demands while meeting QoS (Quality-of-Service) requirements.