Red Hat Enterprise MRG Benefits

Red Hat Enterprise MRG provides a unified, high performance distributed computing platform by integrating the capabilities of MRG Messaging, MRG Realtime, and MRG Grid.

  • What if you could manage demanding workloads using high performance computing (HPC), high throughput computing (HTC), and Grid computing?
  • What if your tasks could harvest idle capacity from the computers on your network, including desktop computers?
  • What if you could take your tasks and run them on local grids, on desktops, or on rented capacity?
  • What if your messaging system could offer both reliability and performance orders of magnitude better than current commercial systems?
  • What if your messaging system was open, standards-based, and interoperable across languages and platforms?
  • What if you could guarantee low, predictable realtime latency for the tasks you run?
  • What if your systems could easily be optimized for reliability, throughput, or latency?
  • What if you could do all of this with one platform, with one set of management tools?

Unprecedented Performance

Red Hat Enterprise MRG offers the highest performance for a variety of distributed computing tasks. Notably:

  • MRG Messaging includes a new high-performance journal tailored specifically for Red Hat Enterprise Linux's I/O system. This enables MRG to provide sustained durable messaging throughput of 500,000 messages/second per LUN, which is orders of magnitude better than previously possible.
  • MRG Messaging provides infiniband support for scaling to unparalleled throughput/latency ratios for transient messages.
  • MRG Realtime provides deterministic low-latency for tasks ranging from messaging to workload scheduling to custom enterprise compute jobs.
  • MRG Grid provides High Throughput Computing (HTC), managing pools of tens of thousands of nodes and hundreds of thousands of jobs.

Architecture Flexibility and Freedom

Red Hat Enterprise MRG is fully configurable, allowing systems to be optimized for performance, throughput, reliability, or other characteristics. System designers and architects can choose what they want to optimize across a variety of distributed computing tasks.

By providing fully optimized distributed computing capabilities in a single integrated platform, MRG lets architects create systems that leverage all the computing power in an organization, without needing to rely on multiple point products to cover tasks like reliable messaging, high-speed messaging, large file messaging, or large-scale grid computing - each with its own platform requirements and management tools. Using MRG for distributed computing allows enterprises to :

  • Achieve significantly higher peak computing capabilities and utilization rates from their IT infrastructure by not having to build dedicated, functional silos.
  • Simplify deployment and management with unified tools.
  • Standardize and simplify computing hardware and software infrastructure.

Interoperability and Deployment Freedom

Red Hat Enterprise MRG provides full interoperability with a variety of platforms and languages, including: Windows and Solaris; Java, JMS, .NET, C++, Ruby, and Python. This enables integration with legacy systems and applications as well as flexibility in developing new applications.

By fully supporting AMQP, MRG is fully inter-operable and interchangeable down to the wire level with other messaging products that support AMQP.

As a fully open source product (licensed under GPL v2), MRG provides transparency and freedom from lock-in.

Revolutionary Synergies

By providing a holistic platform for higher performance distributed computing, MRG provides many new capabilities for enterprise computing, including:

  • Flexible workload processing - depending on latency, performance, and throughput needs, a given workload might be processed under control of MRG Realtime to provide low, predictable latency for system calls, or using MRG Grid for High Throughput Computing or High Performance Computing.
  • Network support for in-flight transaction re-prioritization. As a founding member of the AMQP working group, Red Hat is working to create a broad ecosystem around AMQP so that even hardware devices like network switches and routers can natively parse and process messages from MRG. This will enable, for example, stock traders to issue cancel orders that overtake already-issued buy orders at the network level through message prioritization in the event of a market crash.
  • Full transactional messaging capabilities through light-weight scripting clients. MRG's 50k python client messaging provides all the functional and transactional power of its Java, .NET, and C++ clients. This provides quick and powerful capabilities for interacting with a messaging system under high-pressure situations.

Future versions of MRG will also offer fast scheduling with realtime-enabled QoS of message-queue-submitted workloads. This means that if a grid job requires high priority and only takes a short time to complete, MRG's scheduler can ensure that task is scheduled and run quickly and reliably. And applications written in a variety of languages can schedule jobs natively by simply sending an AMQP message to MRG's scheduler.

Messaging, Realtime, and Grid

In addition to the benefits of MRG as a distributed computing platform, it also has best-of-breed capabilities in the focused areas of messaging, realtime, and grid. For more information on these benefits, visit the detailed pages around the technologies: