Scala Computing has developed a novel neural-net based job scheduler ideal for demanding, mission critical HPC applications. Oasis provides users with a comprehensive set of scheduling features that optimize for fairness policy as well as maximum system utilization. The result is decreased time to results, shorter queuing time for end users, and enhanced performance at scale. Scala Oasis is ideal for users who want an intelligent job scheduler that drastically reduces time to results for computationally intensive jobs.

Neural net based scheduling

By leveraging the power of machine learning, Scala Oasis will continuously improve its scheduling policy. This means that the more jobs you run, the more Oasis learns, decreasing time to results and cutting operational expenses. Oasis is the ideal job scheduler for enterprises that want to optimize their hardware specifications to ensure decreased time to results and reduced hardware costs.

Maximum resource utilization

Scala Oasis maximizes the use of your available compute resources by having contextually aware scheduling policy. All jobs submitted to Oasis will be allocated to their optimal cluster - this translates to fewer idle compute resources, regardless of environment. In addition, Oasis can be configured to your organization’s custom policy requirements; we can train Oasis to schedule in a manner that best fits your organization’s demands (e.g. prioritizing energy efficiency, etc.).

Increased computational capacity

Decreased time to results and queuing time means that you can submit more jobs to Oasis than on other platforms. Likewise, with Scala’s proprietary Job Scheduler Module (JSM) technology, Oasis finds an optimal allocation of compute resources with regards to network latency, meaning that you can submit more computationally complex jobs to the same environment without investing in additional hardware. Together, Oasis delivers significantly increased computational capacity and performance, for enterprises that deal with high volumes of HPC jobs, highly complex applications, or both of the above.