Slurm

Why Adaptive Computing with Slurm?

We have developed a number of powerful add-on’s with the ability to expand the foundation provided by Slurm. These features include the Viewpoint job submission portal, Remote Visualization, Reporting & Analytics, Nitro High Throughput, and Accounting. While these modules were originally created for Moab, we have added cross-scheduler support for our friends on Slurm, so that you can take advantage of these value-adds without having to make large changes to your scheduling environment (some professional services may be required). We hope we can provide additional value to your computing systems!

Viewpoint

Viewpoint is an easy-to-use job submission and management GUI designed to unify and simplify the submission experience for users and administrators. Viewpoint’s ease-of-use makes it possible to expand an HPC user base to include even non-IT skilled personnel, in addition to increasing the productivity of every worker.

  • Job Submission – Submit and manage basic jobs through a simple and easy-to-use point-and-click web portal, in addition to the command line interface.
  • Application Templates – Automate best-practices by pre-defining default values, hiding unnecessary options, and adding custom fields in application templates, thereby simplifying the submission process, reducing errors, and optimizing application run time.
  • Script Builder – facilitate correct use of syntax and reduce new-to-HPC user error through a visual script builder. Drag and drop accepted parameters, and directly incorporate application template-defined parameters.
  • Job Details – Gain insight into job status and reasons for failure through explanatory messages, saving administrator time by helping users help themselves.
  • File Manager – Navigate output files right from an internal browser that includes the ability to move, rename, delete, compress, uncompress, and preview files.

Remote Visualization

With remote visualization, users spend less time waiting for files to be moved between their workstation and the cluster as they simply render remotely and visualize it locally. Users also save time due to more effective collaboration through shared sessions. With many skilled workers being paid over $US 100/hour, it quickly pays for itself through productivity savings.

Organizations will avoid having to purchase high-end desktops for all their knowledge workers, instead sharing systems with expensive licenses or GPUs. They also avoid overbuying on their network, since they will have reduced much of the file transfer requirements between the desktops and back end servers.

  • Full Desktop or Single Application – Display full-size Linux/Unix desktops or a single application in a rootless window on a Windows or Mac desktop.
  • Pixel Perfect – Users are able to experience pixel-perfect color & rendering.
  • Universal Keyboard Selection – Automatically configures to the right keyboard for any language.
  • Copy/Paste – Supports copy/paste of data to and from the remote Linux application and Windows/Mac applications on the PC.
  • Dynamic Display Resizing – The user experience includes a unique ability to dynamically resize the remote Linux desktop on their desktop.
  • Multiple Screens and Displays – Supports the ability to spread the display across multiple monitors or to have multiple simultaneous X Displays.
  • Suspend/Resume – Users can suspend their sessions, disconnect their PC and resume the session at a later time from any device.
  • OpenGL/3D Rendering – Supports the remote rendering of OpenGL-based Linux and Unix applications.

Reporting & Analytics

Adaptive Computing’s Reporting and Analytics tool enables organizations to gain insights by streaming resource usage and workload data into custom reports and personalized dashboards. Adaptive Computing’s Reporting and Analytics tool enables organizations to stream in and organize their data to gain insights to help drive better decision-making and policy enforcement. This drives improved resource utilization and efficiency, better capacity planning, and greater alignment of resources to mission objectives.

Integrated directly with Viewpoint, this tool becomes a natural empowering extension for administrators and managers. The Reporting and Analytics engine comes with dozens of pre-defined reports as well as a rich UI-based designer which allows for customization.

Example Out-of-the-Box Reports:

  • Average Wait Times – User, group, account, class & QoS
  • Job State Duration – Idle, Starting, Running, Hold, etc.
  • Highest Requested – Resources, partitions, features, etc.
  • Allocated/Utilized Resources – User, group, etc.
  • Outage Impact – Node state %’s per time period
  • Job Start Time Delay – Identify wait times
  • Core Use Efficiency – Per utilized node

Nitro High Throughput

Nitro is a highly powerful, yet simple task launching solution which operates as an independent product but can also integrate seamlessly with any HPC scheduler. Instead of requiring individual job scheduling, Nitro enables high-speed throughput on short computing jobs by scheduling only once for a large set of jobs.

  • Enables 100x faster throughput on short jobs
  • Supports thousands to billions of tasks in HTC or SOA workloads
  • Removes launch speed bottleneck and achieves significant improvement in overall system efficiency

Accounting

As users and organizations receive feedback about their usage or even get charged for it, their usage behavior inevitably changes to drive improved overall efficiency. Moab Accounting Manager (MAM) is a fast, easy-to-use accounting management system that allows showback and chargeback for usage tracking and charging for resource or service usage in virtually any computing environment.

  • Track usage per user, group, project, or account
  • Pre-pay or pay-as-you-go models
  • Charge rates per resource, action, or quality of service
  • Lien-based model
  • Enforce budgets