Gigantic clusters: where are they and what are they doing

C lusters are like skyscrapers—mammoth structures towering above their smaller counterparts—except that skyscrapers have longer lives. The race to build ever bigger clusters is on, largely driven by emerging scientific and engineering applications. Several research efforts are underway at various universities and US research laboratories. In this article, I examine some of the largest clusters in the world, providing some recent news and opinions from the experts. A cluster is a collection of complete computers (nodes) interconnected by a high-speed network. Typically, each node is a workstation, PC, or symmetric multiprocessor (SMP). Cluster nodes work collectively as a single computing resource and fill the conventional role of using each node as an independent machine. A cluster computing system is a compromise between a massively parallel processing system and a distributed system. An MPP system node typically cannot serve as a standalone computer; a cluster node usually contains its own disk and a complete operating system, and therefore, also can handle interactive jobs. In a distributed system, nodes can serve only as individual resources while a cluster presents a single system image to the user. In recent years, the performance of commodity off-the-shelf components, such as processor, memory, disk, and networking technology, has improved tremendously. Free operating systems, such as Linux, are available and well-supported. Several industry-standard parallel programming environments, such as MPI, PVM, and BSP, are also available for, and are well-suited to, building clusters at considerably lower costs than their counterpart MPP systems. Scalability—a system’s ability to scale its resources and sustain its performance accordingly—is fundamental to parallel computing. Although this concept has been realized in MPP systems (both nonshared and shared memory architectures), it has been less certain with clusters, largely because of the software and hardware overheads in the interconnection networks. Integration of all the system resources, such as processors, memory, disks, networking, and I/O subsystems, is an important factor. This situation is now changing, and considerably larger clusters (in excess of 32) are being built. Some questions that arise are