Running in the cloud is tremendous fun to talk about. However, let's remember that the cloud is, for all intents and purposes, just “someone else's computer”. Furthermore, going to the cloud does not de facto mean that you will realize performance gains. I would like to take a brief moment to summarize 3 recommendations to improve the speed of your cloud simulation.
-
Choose a purpose-built cloud simulation platform that is optimized for your workload.
-
Leverage HPCaaS to eliminate headaches.
-
Start with a hybrid HPC approach and ease into on-demand CAE.
Deciding to run simulations in the cloud is a great first step. Arriving at the maturity to run CAE smoothly in the cloud is almost impossible with a DIY approach. Working within an ecosystem tuned for your engineering workload will ensure that all the right knobs are turned for successful implementation.
Now that we have established it is most productive to bring in HPC experts, we should address the optimal business model to consume those services. High Performance Computing as a Service (HPCaaS) provides all the benefits of starting quickly as well as regular updates to the HPC environment to stay current with the latest CAE version. Gone are the days of troubleshooting upgrades, libraries, and license servers.
Jumping into the cloud all at once can be amazing… and scary. For those more risk-averse, it makes sense to ease into the cloud with a hybrid HPC approach. This method leverages existing on-premise hardware in addition to the cloud. One version of this scenario is bursting to the cloud for especially large simulations. A second version is splitting up simulations to be either onsite or cloud-based, determined by their level of confidentiality. A third version is splitting up simulations by the user group running the jobs. Once the team is comfortable, then migrating to completely on-demand HPC will be a natural progression.
Contact us to learn more about HPCaaS and Hybrid Cloud solutions.
