Dynamically generating pre-aggregated datasets

A non-transitory computer readable medium storing instructions that cause one or more data processing apparatus to perform operations comprising:

• receiving one or more requests for a first type of performance measure;
• identifying, based on the one or more requests and from a dataset, a proper subset of data used to determine a value for the first type of performance measure;
• determining the value based on the proper subset of data;
• determining a first latency specifying an amount of time that was required to determine the value using the dataset;
• defining a pre-aggregated dataset that includes the proper subset of data from the dataset;
• calculating a benefit score for the pre-aggregated dataset based on a difference between the first latency and a second latency specifying an amount of time required to respond to the request using the pre-aggregated dataset;
• adjusting the benefit score as a rate of received queries for the first type of performance measure changes over time;
• determining, at a given time, that the adjusted benefit score meets a threshold value;
• and
• storing the pre-aggregated dataset in response to determining that the benefit score meets the threshold value.

A system, comprising:

• a data store storing a dataset associated with one or more types of performance measures;
• and
• one or more processors that interact with the data store and execute instructions that cause the one or more data processing apparatus to perform operations comprising:
• receiving one or more requests for a first type of performance measure;
• identifying, based on the one or more requests and from a dataset, a proper subset of data used to determine a value for the first type of performance measure;
• determining the value based on the proper subset of data;
• determining a first latency specifying an amount of time that was required to determine the value using the dataset;
• defining a pre-aggregated dataset that includes the proper subset of data from the dataset;
• calculating a benefit score for the pre-aggregated dataset based on a difference between the first latency and a second latency specifying an amount of time required to respond to the request using the pre-aggregated dataset;
• adjusting the benefit score as a rate of received queries for the first type of performance measure changes over time;
• determining, at a given time, that the adjusted benefit score meets a threshold value;
• and
• storing the pre-aggregated dataset in response to determining that the benefit score meets the threshold value.