The Power of Clusters
Some applications are able to run well on a single computer, however, critical applications required for a successful enterprise need more than that. Your high-demand applications (also referred to as jobs in cluster computing) benefit from running on a cluster if they're in need of simultaneous access to many CPUs, disk drives, or both.
A cluster is essentially a group of computers capable of running an application in parallel across many devices. The ultimate goal of clusters is making your critical applications run fast enough to be useful. Windows HPC Server 2008 is a perfect solution for cluster-based operation, targeted towards large businesses and organizations looking to build high computing environments.
To match the requirements of successful cluster management, the Administration Console has received some changes. It is now a suitable way to manage every aspect of your cluster management. The changes were based on the System Center UI framework, which integrates every action you need to keep your clusters under control.
Job scheduling is one of the two primary functions of the head node software found in Windows HPC Server 2008. It fulfills functions such as accepting jobs submitted by users, assigning jobs to compute nodes, and monitoring job progress over time. Cluster administrators can specify new scheduling policies, job priorities, and many other factors to maximize performance. Job scheduling now comes with support for external databases, as well as granularity at processor core, socket, and compute node levels.
Your administrators can now utilize node grouping in order to categorize and batch operations on compute nodes. This saves a lot of precious time, maximizing the efficiency of nodes.
Get critical information at a glance with the new heat map for your clusters. See the status of your clusters at a glance with the heat map interface which displays a visual view of what needs attention. Stay on top of your business at any time without having to dig for information, or go through long hours of troubleshooting.
One of the benefits of HPC solutions includes the ability to use parallel computing. This is achieved through integration with WCF (Windows Communication Foundation) allowing SOA application developers to take advantage of the power offered by parallel computing.
Parallel computing allows you to carry out a huge number of calculations and execute many processes simultaneously. Divide large problems into smaller ones, and solve everything at the same time more efficiently. The main benefit of parallel computing is the ability to execute code efficiently, saving you a huge amount of time.
Windows HPC Server 2008 comes with multiple built-in features that allow you to work more efficiently with clusters. This includes built-in support for cluster-wide updating, system diagnostics, and cluster reporting.
NetworkDirect is a new RDMA (Remote Direct Memory Access) network interface that's included in Windows HPC Server 2008. Benefits of NetworkDirect include robust improvements for the performance of MPI applications running over high-speed fabrics.