Windows HPC Server 2008 R2 Supports Interoperability with Linux in Dual-Boot and Hybrid HPC Environments
Have it both ways.
Dual-Boot and Hybrid clusters provide a new model for getting greater productivity from the investment an organization makes in HPC hardware.
Benefits of Dual-Boot and Hybrid Clusters:
Make it easier for all parts of your organization to benefit from your HPC resources by providing operating system flexibility
Expand your HPC investments to new users by providing greater application and platform choices
Allows Windows and Linux jobs to be optimally scheduled across compute nodes, based on the OS mix that is currently available on the cluster
Ready to respond to increased demand for applications and users on either platform
More easily add Windows HPC capability to existing Linux systems without having to make an “either-or” decision
Head nodes remain independent, and jobs are scheduled separately to each head node
Read the Multiple-cluster shared infrastructure white paper »
Common Submission Portal and/or Queue
Hybrid clusters manage the job queues for both Windows and Linux HPC as a single job submission queue. The hybrid cluster model also centralizes cluster reporting, SLA policies, and accounting for both Linux and Windows HPC into a single solution.
Dual-boot clusters are setup to more easily switch the operating system on compute nodes based on any number of mechanisms including manually, time of day, automated scripting, job load, and many others. The Windows HPC and Linux job schedulers schedule jobs in their independent queues based on the currently available compute nodes.
Dual-boot systems allow users and administrators to treat both the Windows HPC and Linux clusters completely separately, with separate job submission mechanisms, reporting, and management, allowing the unique benefits of each solution to shine. Review the whitepapers below for more information. We've also partnered with Clustercorp and IBM on dual boot solutions.
Clustercorp’s Rocks+Hybrid
Microsoft has partnered with Clustercorp to create Rocks+Hybrid, a complete HPC cluster software stack, including everything from the operating system to a modular applications-level environment. Rocks+ is deployed as a single DVD to create a turnkey solution that is simple for hardware vendors or end-users to deploy.
Read more »
IBM Extreme Cloud Administration Toolkit (xCAT)
xCAT is designed to automate most of the deployment task necessary to bring a variety of operating sysetms up and running on bare metal. This makes it particularly good at dual-booting clusters between Linux and Windows HPC, with its unique distinction that is supports diskless nodes in a dual-boot configuration as well.
Read more »
Hybrid Server Cluster Architectures
Hybrid cluster designs provide users with more dynamic resource automation and common job submission capabilities.
Hybrid clusters combine the job queues for both Windows and Linux HPC into a single submission queue. This allows hybrid clusters to dynamically react and update the cluster configuration based on the jobs in the actual queue. The hybrid cluster model also centralizes all of the cluster reporting and administration for both Linux and Windows HPC into a single solution. Hybrid Clusters:
Provide the ability for a more real-time cluster configuration response based on actual job workloads or historical job submission patterns
Provide a single job submission mechanism for both Windows and Linux HPC jobs
Allow Windows and Linux jobs to be real-time balanced across compute nodes, regardless of which OS a node happens to be running
Allows organizations to provide service level agreements for workloads that span Windows and Linux HPC clusters
Automate business processes that include both Windows and Linux HPC workloads
Hybrid Cluster Partner Solutions
MOAB Adaptive Computing
As an intelligent metascheduler, Moab determines when the OS mix should be modified for optimal efficiency, based on defined policies and service-level agreements and on current and projected workload. When specified conditions are met, Moab automatically triggers the OS change using a site’s preferred OS-modification technology, such as dual boot, diskful, or stateless (diskfree) provisioning.
Platform Computing
Platform ISF Adaptive Cluster turns static clusters into dynamic, shared computing environments using heterogeneous physical and virtual HPC resources. It allocates resources dynamically based on Platform LSF and Windows HPC Server workload demands.
