Hardware configuration file: hardware.json

This JSON file contains parameters for different hardware configurations. Only one configuration is used at a time. Each root level key is a different hardware configuration name and this contains a dictionary of hardware configuration parameter key-value pairs.

Hardware Configurations

The following hardware configurations represent a cost split ratio between compute nodes and total storage. Other configurations can be added as needed.

  • 50_50: Equal split between compute nodes and total storage (800 nodes, 500 TB performance storage, 35 PB capacity storage)

  • 75_25: A 75/25 split between compute nodes and total storage (1200 nodes, 250 TB performance storage, 17.5 PB capacity storage)

  • 25_75: A 25/75 split (400 nodes, 750 TB performance storage, 52.5 PB capacity storage)

  • control: A control configuration for simulation experiments with 1e6 nodes, 1e6 TB performance storage, and 1e6 PB capacity storage. This is the default configuration.

Parameters

  • capacity_storage_pb: Total capacity storage, in petabytes, for holding (calibrated) visibilities and data products

  • performance_storage_tb: Total performance storage, in terabytes, for holding intermediate data files produced by batch processing

  • compute_nodes: Number of compute nodes available for batch pre-processing

  • data_product_storage_pb: Total data product storage, in petabytes, for holding data products produced by the pipeline

Note

If the hardware configuration is not sufficient for any one of the scheduling block instances the simulation will fail.