United States   Change   |   All Microsoft Sites

Home

Previous Version

Microsoft is taking HPC further with Windows HPC Server 2008, which was built on the strengths of its predecessor, Windows Compute Cluster Server 2003 (WCCS). Windows HPC Server 2008 provides enhanced productivity, scalable performance, and ease of use with a new scheduler, better provisioning, and a new management interface.

Please use the following as a guide to feature comparisons between the two versions.

Feature 

Windows Compute Cluster Server 2003 

Windows HPC Server 2008

Operating System Support 

Windows Server 2003 SP1 

Windows Server 2008 HPC Edition, Standard (64-bit), Enterprise (64-bit), Datacenter

Processor Type Support 

x64 (AMD64 or Intel EM64T) 

x64 (AMD64 or Intel EM64T)

Maximum number of sockets 

4 CPUs 

4 CPUs (HPC Edition)

Maximum amount of memory 

32 GB of RAM 

128 GB of RAM (HPC Edition)

Systems Administration

User-friendly interface of Administrator Console enabled simpler completion of essential tasks such as configuring cluster networking topology, setting user permissions, and monitoring jobs.

Clusrun allows cluster administrators to submit a single-task job which bypasses the job queue and is immediately executed on a specified set of compute nodes.

Cluscfg configuration commands.

Remote Installation Services used when deploying nodes with the automatic method.

New Administrator Console user interface based on System Center. The new Administrator Console integrates all aspects of cluster management and simplifies completion of essential tasks such as configuring cluster networking topology, setting user permissions, and monitoring jobs and the operational health of the cluster.

Clusrun UI is improved and now allows administrators to run scripts in parallel across the cluster.

New features such as Node Templates leverage Windows Deployment Services (WDS) to significantly improve compute node deployment. Support for WDS replaces the existing RIS technology for node provisioning.

Cluster administrators can create Node Groups to view and manage collections of nodes.

Systems Administration includes a newly integrated heat map, which provides an at-a-glance view of cluster-wide or node group status, performance, and utilization.

New cluster and node diagnostic testing allows a systems administrator to store, filter, and view test results and history.

Built-in reporting integrated into management console.

All admin tasks are now accessible via PowerShell.

Head node high availability support through Windows Cluster Services.

End User Tools

Command-line interface.

User console made submission and tracking of user jobs simple.

PowerShell commands for scheduling and managing jobs. Windows HPC Server has full compatibility with existing Compute Cluster Server 2003 commands and also offers full support of PowerShell. More than 130 command-line tools enable automation of system administration tasks.

New Job Manager interface with improved support for parametric commands.

Custom job view filtering.

Networking

Wizard-driven configuration of network services such as DHCP, ICS, RIS, and the firewall.

Wizard-driven configuration of network services such as DHCP, RRAS, DNS, and firewall.

Support for NetworkDirect, a new RDMA networking interface built for speed and stability when using high speed interconnects.

MPI

Security-enabled Windows programming interface for MPICH-2 integrated with Windows Active Directory.

Security-enabled Windows programming interface for MPICH-2 integrated with Windows Active Directory.

Integrated with new RDMA networking interface NetworkDirect.

MS-MPI events are integrated with Event Tracing for Windows.

New shared memory communication implementation.

Job Scheduling and Job Policies

Support for batch or MPI-based jobs.

Job Policies:
* "First-come, first-served" scheduling
* Backfilling of jobs
* Exclusive scheduling
* License scheduling

Job Templates for setting constraints/requirements on jobs, including the ability to route jobs to specific sets of resources. Job Templates also simplify end-user job submission by enabling cluster administrators to specify default job term values.

New Job Policies:
* Resource Matching
* Multilevel resource allocation; resources include nodes, processors, and sockets
* Adaptive allocation allows the growing and shrinking of running jobs
* Pre-emption of running jobs

New compute nodes designated as WCF Brokers allow interactive jobs to occur between clients and compute nodes. WCF Brokers can be added to support increased job capacity for SOA jobs.

Support for the Open Grid Forum HPC Profile for Job Scheduler interoperability.

Developer Tools

C# and COM API for the CCS 2003 Job Scheduler.

MS-MPI.

Support for interactive Service-Oriented Applications (SOA) using the Windows Communication Foundation (WCF).

Scalable Microsoft HPC Pack 2008 API for job scheduling integration.

Ability to export ETW tracing into jumpshot format for third-party HPC tools support.

Support for all WCCS COM APIs.

 

For further information about Windows Compute Cluster Server 2003, please visit the WCCS Technical Library.

Product Information