Considerations for HPC and AI in the Cloud

HPC enables applications to run at greater performance and on larger problem sizes, and for two decades has been implemented almost exclusively as tightly integrated clusters comprising commodity technologies and special software stacks. Cloud computing has emerged over the past decade as a dominant method of providing computational and storage resources for consumer and enterprise applications, and has increasingly leveraged most of the same technologies used in high performance computing clusters, albeit with different usage and payment models for different workloads and applications. However, the many benefits of cloud computing--rapid access to large amounts of resources, application abstraction from specific hardware, flexible payment models including consumption-based usage, user-based request and provisioning, and more--on top of these large pools of hardware have rapidly become attractive to HPC users and customers. Cloud computing also offers a choice in models — including infrastructure as a service (IaaS), platform as a service (PaaS), software as a service (SaaS) and managed services — and HPC users continue to integrate cloud computing resources into their workloads. As more and different types of users rush to leverage advanced computing resources, it’s important to consider the pros and cons of cloud computing. In this session, HPC veteran Jay Boisseau will discuss the top considerations for HPC, AI and high performance data analytics (HPDA) in different types of clouds.