Record sampling

Record sampling refers to the ability of policy monitoring to randomly select samples of entity records from the operational server and to load the records into the landing area.

Record sampling helps to reduce the time that is required to extract and process full data sets, which can include many millions of records, from the operational server.

• If the number of entity records is below 1 million, then the full data set in the operational server should be processed and loaded into the landing area.
• If the total number of entity records is significantly greater than one 1 million then, you should use record sampling.

In general, the larger the number of records used for the sample, the closer the results of the sample will be to the results of a full run.

Typically, a randomly selected set of 1 million records yields KPIs with a deviation within 1%.