High Performance Dynamic Clusters with IBM Platform Dynamic Cluster
Many high performance computing (HPC) environments suffer from low utilization because the workload they process is heterogeneous and dynamic, but the clusters they run on are static. IBM Platform Dynamic Cluster turns static clusters into dynamic, shared cloud resources.
The price of siloed clusters and static resources
To meet application and user requirements, many Technical Computing environments have implemented costly, inflexible silos. These environments are often inefficient with less than fully utilized hardware; they require a significant amount administrator time to install, monitor and manage. The time it takes to obtain results depends on the time and availability of people to execute technical tasks such as reprovisioning operating systems based on workload requirements.
Workload-aware dynamic computing
A dynamic, shared cloud environment can dramatically improve service levels and resource utilization. IBM® Platform™ Dynamic Cluster turns static IBM Platform LSF® clusters into dynamic, shared cloud resources. Because it can automatically change the composition of clusters to meet ever-changing workload demands, you enjoy improved service levels and shorter time to results. Platform Dynamic Cluster makes it possible to do more work with less infrastructure.
Unlike many similar solutions in the market, Platform Dynamic Cluster has the flexibility to automatically provision mixed physical and virtual environments. It leverages your existing investments in hypervisors, management tools and virtual machine (VM) templates to create a dynamic private cloud environment. With intelligent policies and features such as live job migration and automated checkpoint-restart, Platform Dynamic Cluster helps you increase utilization, while reducing administrator workload.
Increase utilization, decrease time to results
Efficient resource scheduling and provisioning are essential to realizing the productivity gains and savings possible with a cloud implementation. Platform Dynamic Cluster can automatically provision an almost unlimited set of operating system and application environments, adapting the cluster to meet changing workload demands in order to:
Maximum efficiency in HPC cloud management
The efficient scheduling and provisioning of resources are essential to realizing the productivity gains and savings possible with a cloud implementation. Platform Dynamic Cluster incorporates unique features that ensure efficient workload placement, taking into account application resource requirements and the state of the cluster. The ability to intelligently reuse virtual machines for batch jobs with similar resource requirements avoids excessive stopping and starting of VMs, keeping throughput high and latency low. In environments with high demand for different types of resources, configuration settings can control the rate of automated node switching, ensuring that unnecessary changes do not inhibit performance and utilization. These and other Platform Dynamic Cluster features combine to maximize throughput and resource utilization, resulting in the highest possible levels of productivity and efficiency.