CTSS 4 Distributed Parallel Application Support Implementation
From TeraGrid Wiki
This capability kit implementation was developed in the GIG Software Integration Area based on the contents of CTSS 3 and our current understanding of requirements from applications and operations communities. It defines an optional capability set for Teragrid resources beginning with CTSS 4. This implementation includes one new capabilities that was not included in CTSS 3, so resource providers and any other TeraGrid service providers will need to deploy that new capability to be compliant with CTSS 4.
The purpose and specific capabilities provided by this kit are described in the companion document, CTSS 4 Distributed Parallel Application Support Capability.
Contents |
Abstract
This document describes the recommended implementation for CTSS 4 Distributed Parallel Application Support Capability. The Distributed Parallel Application Support kit defines a capability that allows scientists to build and run parallel applications across multiple resources.
The purpose and specific capabilities provided by this kit are described in the companion document, CTSS 4 Distributed Parallel Application Support Capability.
Implementation
The information services name for this kit is distributed-parallel-app.teragrid.org.
Components That Do Not Change From CTSS 3 to CTSS 4
Resources that have these CTSS 3 components already installed do not need to re-install them.
Parallel Application Support
| Package name: | RP determined distributed parallel tools/libraries |
|---|---|
| Package version: | Implementation determined |
| Software components: | Tools for compiling and running parallel applications |
| Software pre-requisites: | Implementation determined |
- One or more distributed parallel application development and execution tools. The Deployment section below lists some components that may be available on TeraGrid resources that satisfy the requirements of this kit. If you offer your TeraGrid users with access to some other distributed parallel application support tool that's not on this list please contact gig-pack. We will jointly determine a naming convention that can be used to register that component in uniform way accross TeraGrid resources.
New and/or Changed Components
Information Services
| Package name: | ctss-distributed-parallel-app-registration |
|---|---|
| Package version: | 4.2.0 |
| Software components: | ctss-distributed-parallel-app-registration |
| Software pre-requisites: | Perl |
- The component includes the information needed for the Distributed Parallel Application Support kit to register to the CTSS 4 Kit Registry information service.
Deployment
Software Deployment
This section describes the steps to be performed to install or update the CTSS 4 components in this kit.
Components that come from CTSS 3 that are already installed, do not need to be re-installed.
| Step | Component | Description | Available At |
|---|---|---|---|
| Required existing CTSS 3 components. Have one or more of these components available to users. | |||
| 1. | mpich-g2 | mpich-g2 from CTSS 3. This can exist with multiple underlying local MPI implementations | CTSS v3 repository |
| 1. | mpiG | mpiG | mpiG Homepage download |
| Required new CTSS 4 components. | |||
| 2. | ctss-parallel-app-registration | TeraGrid Parallel Application Support Kit registration | CTSS v4 repository |
SoftEnv Configuration
This section lists the SoftEnv keys, macros, and associated descriptions to define for components in this kit.
- Version numbers below are examples and may be changed to match what's installed and allowed above.
- Define release keys for all deployed releases.
- Define version keys pointing to recommended release key.
- Define component keys pointing to the recommended release key.
- Only register version keys in information services.
| Key | Description | Notes |
|---|---|---|
| @mpich-g2-1.2.6e-mpich-gm-intel81 | mpich-g2 v 1.2.6e using Myrinet, Intel 8.1 | |
| @mpich-g2-1.2.6e-mvapich-intel91 | mpich-g2 v 1.2.6e using mvapich, Intel 8.1 |
Security Considerations
Resources Required *
GIG Resources
The following GIG resources will be required to prepare this kit for deployment and to maintain it during its operational lifespan.
Software Integration
Documentation
User Services
Science Gateways
Operations
Security
RP Resources
Resource providers who choose to implement this kit on their systems will incur the following resource requirements.
Software Deployment
Software Configuration
User Support
Maintenance
System Load
Scaling *
Verification & Validation
The TeraGrid operations team will need to configure Inca to run the appropriate set of Inca tests for any resources that choose to implement this capability kit. The data in the CTSS software and service registry may be used to automatically configure Inca to run these tests, or the configuration could be performed manually.
Documentation
The TeraGrid documentation team will need to ensure that the TeraGrid documentation (User Info pages, User Portal as appropriate) reflects the availability if this capability kit on any resources that choose to implement this capability kit. The data in the CTSS software and service registry may be used to automatically list all appropriate resources, or the documentation changes could be made manually.
Acknowledgements *
This work was supported by the National Science Foundation Office of Cyberinfrastructure, grant number 0503697 ETF Grid Infrastructure Group: Providing System Management and Integration for the TeraGrid.
Author Information *
Eric Blau
John-Paul Navarro
University of Chicago / Argonne National Laboratory
9700 S. Cass Avenue
Argonne, IL 60439
blau@mcs.anl.gov
+1 630-252-5638
navarro@mcs.anl.gov
+1 630-252-1233
