Issue link: https://read.uberflip.com/i/1187300
Amazon Web Services – HPC Lens AWS Well-Architected Framework Page 7 the simulation places extra demands on the computing infrastructure, such as the need for high-performance compute nodes and network infrastructure. The infrastructure that is used to run the huge variety of HTC and HPC applications is thus differentiated by the degree of support for coupling of processes across nodes. We present two separate categories of example architectures for HTC and tightly-coupled HPC workloads in this whitepaper. Each workload is presented generically, though callouts are made for variations in scenarios. Traditional, on-premises clusters force a one-size-fits-all approach to the HPC/HTC cluster infrastructure. However, the cloud offers a wide range of possibilities and allows for optimization of performance and cost. In the cloud your configuration can range from a traditional cluster experience with a scheduler and a master node, to a cloud-native architecture with the advantages of cost efficiencies obtainable with cloud-native solutions. Consider the following fundamentals when selecting an HPC infrastructure on AWS: • Network: Network requirements can range from cases with low requirements, such as HTC applications with minimal communication traffic, to tightly coupled and massively parallel applications that require a performant network with large bandwidth and low latency. • Storage: HPC calculations use, create, and move data in unique ways. Storage infrastructure must support these requirements during each step of the calculation. Input data is frequently stored on startup, more data is created and stored while running, and output data is moved to a reservoir location upon run completion. Factors to be considered at each step include data size, media type, transfer speeds, shared access, and storage properties (for example, durability and availability). • Compute: The AWS instance type defines the processor type, speed, accelerator options, and the memory-to-core ratio. Each instance type has access to different network options as well. On AWS, an instance is considered to be the same as an HPC node. These terms will be used interchangeably in this whitepaper. AWS also offers managed services with the ability to access compute without the need to choose the underlying EC2 instance type. AWS