What tools does the XSEDE Cyberinfrastructure Resource Integration team provide to automate the process of building, updating, and maintaining high-performance scientific computing clusters?
The Extreme Science and Engineering Discovery Environment (XSEDE) Cyberinfrastructure Resource Integration (XCRI) group distributes the following tools to simplify the process of building, managing, and updating high-performance computing (HPC) clusters capable of running large-scale science experiments and processing large amounts of data. Using these tools, system administrators and research scientists with relatively little cluster experience can build and manage systems that run the same open source scientific applications installed on XSEDE digital compute services. By offering these tools at no cost to the US open science community, the XCRI team seeks to facilitate computational science research and discovery by providing broader access to high-performance computing resources.
On this page:
The XSEDE Compatible Basic Cluster (XCBC) is a suite of packages designed for use by system administrators who are either building new Linux-based computational science clusters from scratch or performing clean rebuilds of existing clusters. The XCBC suite combines the latest versions of the OpenHPC project and the XSEDE National Integration Toolkit (see below). With a combination of scripts and Ansible playbooks, the XCBC Toolkit allows the quick and easy creation of a new HPC system. It streamlines the process of configuring a "bare- bones" CentOS HPC cluster and minimizes the complexity of expanding its feature set, helping system administrators create computational science clusters comparable to, and compatible with, XSEDE digital services. For more, see What is the XSEDE Compatible Basic Cluster (XCBC)?
On this page:
The XSEDE National Integration Toolkit (XNIT) is a YUM (Yellowdog Updater, Modified) repository designed for use by system administrators who are either building new Linux-based computational science clusters or only needing to install certain scientific applications on existing clusters. The XNIT repository includes open source computational science packages available on larger XSEDE systems. For more, see What is the XSEDE National Integration Toolkit (XNIT), and how do I use it? Additionally, for the full list of XNIT packages, see What packages are included in the basic XCBC and XNIT software suites?
Ansible is an open source platform for automating configuration management, service orchestration, cloud provisioning, and application deployment in multi-tiered IT environments. Ansible is "agentless", using SSH to push changes from a single source to multiple remote resources. Commands can be invoked either ad hoc on the command line or via "playbooks" written in YAML (YAML Ain't Markup Language), a Unicode-based, human-readable, and computationally powerful data serialization language.
The XCRI team provides an Ansible installation script and several Ansible playbooks designed to help deploy full-featured science clusters built with the XSEDE Compatible Basic Cluster (XCBC) suite or the XSEDE National Integration Toolkit (XNIT) repository. Currently, Ansible playbooks are available for installing a Globus Connect server, configuring the TORQUE resource manager, installing and updating XNIT packages, and building an OpenHPC-based cluster. To get the XCRI team's Ansible installation script and playbooks, see the XSEDE XCBC_Ansible software page. For installation instructions, see the Installation page on the Ansible Documentation website.
For more about Ansible, see What is Ansible?