Consider a dataset with a total size of 15GB in memory, as would be returned by the operation df.memory_usage(deep=True).sum().
Running one (1) operation on this dataset, such as a .sum or a .sort_values, would consume 15GB of processed data in Terality.
Terality only charges the data effectively processed by the engine. For example:
* df.head() would only process 5 lines of the Dataframe, not the full Dataframe.
* df.shape won't process any data, and won't be charged by Terality.
Please note that Terality charges per trenches of 1 TB of data processed.
Success-based pricing means that billable usage is only recorded when task runs enter a Success state.