The HPC Basic Profile specifies how to achieve interoperability between job schedulers managing HPC resources located in different administrative domains, enabling common usage scenarios found within many research and commercial organizations.

Example usage scenarios include: Web Application: Users access applications with their web browser while the application server uses the HPC Basic Profile to initiate the execution of the application. The user does not need to be aware of the actual resource manager being provided. Metascheduler: The user submits their job to a local resource manager, which then uses the defined policy to route the job to a suitable execution resource. This resource may be accessed through the HPC Basic Profile and use a different resource manager or be located in a different organization. Rich Client: An engineer or scientist runs a “workbench� application that includes running simulations that test elements of the design. When the user runs a simulation, the rich client dispatches it to a compute resource using the HPC Basic Profile. Workflow Engine: Various applications are chained together into a workflow of computing tasks. Users define job nodes and their dependencies into a graph. The workflow engine does not need to be aware of the different resource managers and can execute each node in the graph based on the job description and the HPC Basic Profile.

The HPC Basic Profile uses Web Services as an interface to expose different resource managers on different operating systems to enable users to submit, monitor and manage jobs on clusters located in different organizations. The HPC Basic Profile is a standard (published as GFD.114) that references existing specifications including:

• OGF Job Submission Description Language (JSDL)
• OGF Basic Execution Service (BES)
• WS-I Basic Profile

In order to run the application, a user requests that a job be run through the resource manager - represented by a compliant HPC Basic Profile web service endpoint. Each application is described using JSDL, an XML-based language for describing the job to be run, while the BES web services defines an interface that allows web services clients to submit, monitor and control jobs through the local resource manager. The HPC Basic Profile defines how the JSDL and BES specifications are composed with components of the WS-Security specification to achieve interoperability.

The HPC Basic Profile leverages standard Web services protocols and development environments. It allows application and middleware software providers to target multiple HPC systems via a single protocol. The demonstrations clearly show how organizations can benefit by

• Integrating commercial and open source products into existing and new HPC systems
• Interfacing with HPC systems at remote locations
• Leveraging generic HPC utilities

Implementations and Interoperability

There are many different implementations of the HPC Basic Profile being developed by organizations involved in the specification process. Many of these implementations are available as open source software, are engineering prototypes of upcoming products or already released products. In order to ensure that these different implementations achieve the primary goal of interoperability, the HPC Basic Profile implementors regularly get together to perform interoperability demonstrations of their software. A common theme of these demonstrations is to be able to show how one would submit jobs from one administrative domain to a cluster residing in a different administrative domain.

These demonstrations have traditionally been held at the Supercomputing conference. The demonstration usually includes the submission of tasks from a client in one booth to a resource manager running in another booth, the retrieval of a task’s execution status and the retrieval of information about an HPC system’s resources.

The most current information about interoperability demonstrations can be viewed at the HPCBP wiki, where the list of participants and HPC Basic Profile implementations are also listed.

About OGF and HPC Profile Working Group

The Open Grid Forum (OGF) is an open community committed to the rapid evolution and adoption of applied distributed computing through activities which explore trends, share best practices and consolidate these best practices into standards. In OGF, people from around the world work together to drive openness and interoperability in the technology and protocols that enable discovery in science and the creation of value in business through applied distributed computing. The HPC Profile Working Group, an OGF group co-chaired by Marty Humphrey of the University of Virginia and Steven Newhouse from EGEE, is part of the compute area within the OGF Standards function.