CloudyCluster GCP Overview
Recognizing that a growing number of people need to use High Performance Computing (HPC) resources, Omnibond designed CloudyCluster to decrease the time and simplify the effort required to create an HPC environment on demand. CloudyCluster allows anyone to quickly set up and configure an HPC environment on Google Cloud Platform (GCP) for HPC parallel workloads.
CloudyCluster makes it easy to set up an HPC environment in the GCP cloud, including scheduler, compute, parallel storage, and native client data access. CloudyCluster and GCP are available 24/7/365, with no hardware and installation downtime. You can pause your environment whenever you are not running a job, so you pay only for the processing time you use and minimal storage fees when your environment is not running.
CloudyCluster quickly provisions compute instances with a wide selection of open HPC applications, dynamically configured for access to storage. All instances within a compute group are based on the same image and the same instance type.
You can create and manage your own CloudyCluster environment using a self-service, mobile-friendly web interface. The CloudyCluster step-by-step wizard makes the setup process a straightforward, painless experience from any device, at your desk or on the go.
Designed with Collaboration in mind, CloudyCluster also supports invitation-based collaboration, providing simple tools to add users and grant permission levels within the basic interface. Collaboration is also supported by simple options to create and share Google folders.
CloudyCluster securly launchings in Virtual Private Cloud (VPC), providing secure parallel computation. With CloudyCluster on Google Cloud Platform, organizations who do not own or have immediate access to large HPC environments can quickly and easily use OrangeFS to make the most of available data whenever needed.
The browser-based CloudyCluster tool creates GCP instances for quick HPC processing. It streamlines the complex GCP setup process for HPC, with options for your unique needs with the following components:
- Parallel storage with OrangeFS
- CCQ, a metascheduler that provides job-initiated elastic HPC
- Schedulers such as Torque and SLURM
- WebDAV for native file access
- Compute instance images contain popular Open Source software packages and can easily be customized
- Globus for site to site data transfers
- All combined by an intuitive web user interface
- Restlike APIs