X
Business

IBM's Intelligent Clusters - an old idea done well

IBM's pre-configured, pre-tested clusters take the uncertainty out of multi-system deployments. Regardless of how good the idea, IBM didn't think of this first. It is yet another legacy of Digital Equipment Corporation.
Written by Dan Kusnetzky, Contributor

I just returned from a fascinating IBM briefing for industry analysts who follow technical computing, the forerunner to what is being called "cloud computing" today. I'm not going to try to discuss everything that was mentioned by each of the excellent presenters. That will come out over time in other commentaries.

Intelligent Clusters

Jeff Falkanger, System x Marketing, spent about 45 minutes presenting IBM's Intelligent Cluster family of products and then facing down a room full of highly technical analysts as they probed IBM's plans for this set of pretested, preconfigured cluster systems, considered the implications and, of course, made many suggestions for how these products could be more successful in the marketplace.

Here's what IBM says about Intelligent Clusters

Highlights

  • Leading-edge technology with flexibility of choice
  • High performance
  • Energy and space efficient
  • Easily deployed, operated and maintained

Leading-edge technology with flexibility of choice

IBM® Intelligent Cluster integrated solutions are built on the highly innovative IBM System x® rack, BladeCenter® and iDataPlex® servers. Whether you are building a small departmental cluster or a super computer, IBM’s broad portfolio of server solutions can be optimized to meet client-specific requirements.

High performance is paramount

High Performance Computing (HPC) workloads will benefit by up to twice the performance. Our integrated cluster solutions offer significant price/performance advantages for many high-performance workloads by harnessing the advantages of highly innovative servers. By design, a range of application environments will benefit, including those optimized for industrial design and manufacturing, financial services, life sciences, oil & gas, government and education.

Energy and space efficient

Intelligent Cluster solutions that integrate BladeCenter or iDataPlex servers can reduce power and cooling costs by up to 50 percent, while maximizing performance density, which is critical for HPC workloads.

Easily deployed, operated and maintained

Intelligent Cluster integrated HPC solutions include servers, storage and industry-leading OEM interconnects that are factory-integrated, fully tested and delivered to your door, ready to plug into your data center, all with a single point of contact for support.

Analysis

It is clear that IBM sees its role in the industry standard systems market as being the go-to company for well-designed, highly integrated and highly open configurations. That being said, the idea to fully configure, test, sell and support highly complex, multi system configurations isn't new. My old Alma mater, Digital Equipment Corporation (now a part of HP), did something very similar in the 1980s by offering pre-configured systems for a number of industries.

IBM, however, has had to take this idea to a new level. Clusters of Industry standard systems are far more complex to configure, test and support because there are many suppliers of systems, memory, storage, networking equipment and software where DEC only had to deal with its own products. This makes IBM's efforts very impressive.

While IBM presents Intelligent Clusters as an offering focused on the needs of the technical computing market, these configurations clearly could be the foundation of in-house clouds, financial services modeling systems, rendering farms for digital content creation, weather forecasting and a number of other very challenging tasks.

I really liked Jeff's suggestion that IBM has learned a great deal through its decades of  "big systems" integration and has taken that learning to build these systems. I believe IBM is correct when it would say their engineering, testing and support capabilities make things work without forcing customers to see these clusters as computer science projects.

Editorial standards