HPC Cloud journey

Understand

Identify HPC workloads
Understand “On-Premise” usage

Prioritize Workloads
Identify attention points
Project costs with CloudSHaper

Prepare

Validate performances
Train your teams
Prototype a new technology

Execute

An unexpected load peak?
A collaboration need?
A recovery plan?

Cluster In a Box

About Cluster-in-a-Box

  • We deploy your HPC clusters on-demand in the AWS cloud in just a few minutes:

    Easy and reproductible launch
    Full visibility and control for IT teams
    Simplicity of use (web interface and remote visualization with EnginFrame and DCV

  • We compose your HPC clusters in the Cloud, according to your needs:

    Select your options (storage, parallel file system, instance type…)
    Benefit from Cloud elasticity, your clusters fit your jobs’ needs
    Benefit from Cloud reliability to reproduce your environments and performance

  • A configurable and intuitive web interface to use your clusters (EnginFrame):

    Secure access
    Job submissions via web forms or CLI
    Selectively publish web submission forms to user groups

  • Your cloud clusters are administered by UCit with adapted SLAs.

  • Follow and control usage and costs through complementary mechanisms:

    Post-analysis of cluster logs with Analyze-IT
    Monthly consumption reports
    Budget consumption alerts (50%, 80%, 90%)
    Auto cluster shutdown below consumption threshold

Technical components

The Cluster-in-a-Box Process

Benefits

  • Variable expenses are indexed on actual consumption
  • Elasticity: only use the capacity you actually need, and adjust upwards or downwards instantaneously
  • You have the choice to access the latest hardware technologies
  • One single point of contact (UCit) for all your technical and administrative needs, with a simple & fast implementation so you can focus on your field of expertise
  • Master your costs thanks to specific financials tools (OKA Financials, OKA Shaper…)
  • Easy to use with an intuitive interface and UCit hosting