ARMONK - 20 Mar 2001: IBM today introduced IBM eServer1 clusters for UNIX2 and Linux3, the world's first systems that tie together a company's entire IT infrastructure -- from Web servers to vital back-end workloads like Enterprise Resource Planning and Customer Relationship Management applications. The new offerings -- originally code-named Blue Hammer -- extend the advanced management software and massive file system of IBM's SP supercomputer to the fast-growing market for rack-mounted servers.
Enabling customers to control up to 32 powerful IBM eServer rack servers from a single workstation, IBM eServer clusters are designed to install, control and maintain huge server farms. The solutions provide maximum performance at an affordable price.
Today's announcement is an important step in IBM's strategy of disseminating the company's best technology throughout the IBM eServer product line. The SP supercomputer is a battle-tested commercial computing platform that has proven its superiority in handling e-business workloads -- such as online stock trading and business intelligence -- for thousands of corporations around the world. Far outclassing cluster software from the competition, the SP's system management software has already been used in real-world data centers to cluster more than 500 separate computers.
"Scaling from one processor to more than 8,100 processors, the SP supercomputer is the most powerful and easy-to-use cluster system ever built, and is at home in a wide spectrum of corporate computing environments," said David Turek, vice president of Deep Computing, IBM Web Servers unit. "By taking the true engine of this system -- the management software -- and bringing it down into rackable UNIX and Linux servers, we are empowering companies to meet the enormous scalability requirements of e-business."
Heart and Soul of the SP
Designed for rack systems, IBM eServer clusters fit perfectly in today's e-business data centers. The Linux version enables clusters of up to 32 two-way IBM eServer x3304 Intel-based machines; while the UNIX version clusters up to 32 two- to eight-way IBM M80 or one- to six-way IBM H80 servers running the AIX operating system.
With IBM eServer clusters, both the midrange M80s and H80s can be clustered with large-scale IBM eServer p680 or S80 servers. The M80, H80, S80 and p680 can also be directly attached to the SP supercomputer.
The heart and soul of IBM eServer clusters are the SP supercomputer's Parallel System Support Programs (PSSP) cluster management software and General Parallel File System (GPFS) software, which were both introduced in 1993. IBM has been honing them ever since -- and both applications reached a milestone recently when they served as an integral part of the official Website for the Sydney Olympics, which received a staggering 11.3 billion hits during the 17 days of the games.
Built upon the system management tools and commands of the AIX operating system, PSSP enables administrators to manage tens or hundreds of clustered servers efficiently and cost effectively. All administrative functions in the cluster, such as operating system installation and configuration, may be performed from a central point of control. With PSSP, operators can monitor and control hardware and keep files synchronized across all servers. It also provides an unsurpassed infrastructure for high-availability, as well as a balanced architecture to ensure that the cluster performs optimally at all times.
GPFS is a highly-available cluster file system, providing shared access to files across all nodes in the cluster. Scaling to nine terabytes, GPFS allows customers to quickly and easily access huge amounts of data on a cluster.
Designed for Mission-Critical Workloads
The IBM UNIX clusters are especially well-suited for consolidation of multiple, diverse applications in large Fortune 1000 companies. Typical applications include Enterprise Resource Planning, Customer Relationship Management, business intelligence, and large-scale Web servers. Clustering enables these enterprises to reduce IT costs, while making their IT infrastructures more responsive to business needs.
The Linux clusters are tailor-made for e-businesses that deploy numerous "rack-and-stack" servers on the fly, as well as firms that need massive processing power to tackle complex problems. These companies include: Internet Service Providers (ISPs) and Application Service Providers (ASPs) that maintain huge server farms; financial services companies looking to consolidate workloads and integrate services; andLife sciences firmsthat need massive computing power to analyze genetic data, design new drugs and solve difficult science problems.
Today's announcement is the second phase in IBM's ongoing efforts to bring sophisticated supercomputing technology to commercial systems. In September 2000, IBM announced the initial Blue Hammer solution, which clusters up to 16 IBM eServer p680 or S80 enterprise UNIX servers running AIX. Future plans include extending Linux clusters to IBM's zSeries and pSeries servers; similarly, UNIX clusters will be extended to additional servers in IBM's pSeries line. IBM also intends to increase the maximum size of its Linux and UNIX clusters to 256 servers in the future.5
System i, System p, System x, System z, BladeCenter, and Supercomputers
Information Management, Lotus, Tivoli, Rational, WebSphere, Open standards, open source