In order to take full advantage of HPC resources, a suite of tools is provided, pre-integrated on the Appliance or within the On-Demand cloud service, that enable engineers to manage job execution, platform utilisation, user priority and solver licence availability and utilisation.
The platform (hardware, software and application environment) itself is supported by the Dezineforce Service Management Agreement, eliminating the need for local HPC management resources and taking care of system performance, upgrades and operation.
Scheduling and workflow management: the Dezineforce Workflow Manager
To maximise efficiency and throughput the integrated Dezineforce HPC Simulation Manager tools manage software licences and hardware compute resources. This enables engineers to submit and execute multiple simulations in parallel, and enables multiple engineers to submit simulations for concurrent or prioritized processing concurrently (Simulation licences permitting).
Hardware scheduling ensures that the compute nodes being used to execute the engineering calculations are loaded efficiently. The Dezineforce scheduler distributes load across the compute cluster, also ensuring software licences are used effectively across all compute clusters.
Dezineforce Workflow Manager works in conjunction with the hardware and licence scheduling to ensure jobs are submitted at the right time and on the right nodes, based on current load conditions and licence availability, job priorities and ‘fairness’.
The Dezineforce Workflow Manager understands the different ways engineering applications report execution errors: some through exit codes, some through output streams, and some by just freezing. The Dezineforce Workflow Manager detects errors and job failures and will restart jobs if it can in order to maintain throughput across the HPC platform.
Smart File System
Engineering processes generate very large data sets rapidly, and in many industries engineers are required to keep design and simulation data for the lifetime of the products (for aviation, this is often several decades). It is also common practice to use a previous design, possibly many years later, as the starting point for a new design. The Dezineforce Smart File System provides the ability to explore stored data without bringing it down to the client, and enables a subset of the data to be loaded, especially useful if long term storage is not available. The Dezineforce Smart File System provides high-performance, secure access with resumable downloads for very large data sets, all available without impacting current simulation or optimisation processing.
Engineer Manager
The Dezineforce Engineer Manager allows the local administrator to grant access rights and job priority to the engineers who are authorised to use the HPC Simulation Appliance for remote solves.
For more information on the Dezineforce HPC Simulation Manager tools contact us.