Summarize records with the stats function

Use the Stats function to perform one or more aggregation calculations on your streaming data. For each aggregation calculation that you want to perform, specify the aggregation functions, the subset of data to perform the calculation on (fields to group by), the timestamp field for windowing, and the output fields for the results.

After the given window time has passed, the stats function outputs the records in your data stream with the user-defined output fields, the fields to group by, and the window length that the aggregations occurred in. The stats function drops all other fields from the record's schema.

The stats function has no concept of wall clock time, and the passage of time is based on the timestamps of incoming records. The Stats function tracks the latest timestamp it received in the stream as the "current" time, and it determines the start and end of windows using this timestamp. Once the difference between the current timestamp and the start timestamp of the current window is greater than the window length, that window is closed and a new window starts. However, since events may arrive out of order, the grace period argument allows the previous window W to remain "open" for a certain period G after its closing timestamp T. Until we receive a record with a timestamp C where C > T + G, any incoming events with timestamp less than T are counted towards the previous window W. See the Stats usage section for more information.

List of aggregation functions

You can use the following aggregation functions within the Stats streaming function:

• average: Calculates the average in a time window.
• count: Counts the number of non-null values in a time window.
• max: Returns the greatest value in a time window.
• min: Returns the lowest value in a time window.
• sum: Returns the sum of values in a time window.

If you have the Streaming ML Plugin installed, you also have the following aggregation functions available:

• etsdc: Calculates an approximate distinct count value for any field.
• perc: Calculates the approximate q-th percentile value of a numeric field.

Count the number of non-null sources per host in a 60 second time window

Suppose you wanted to count the number of times a source appeared in a given time window per host. This example does the following:

• Uses the count aggregation function to count the number of non-null sources and outputs the result to num_events_with_source_field.
• Groups the fields in the output by host.
• Executes the aggregations in a time window of 60 seconds based on the timestamp of your record.

Steps

1. From the Canvas view of your pipeline, click on the + icon and add the Stats function to your pipeline.
2. In the Stats function, add a new Group By.
   1. In Field/Expression, type host.
   2. Click OK.
3. In the Timestamp field, type timestamp.
4. In the Window length field, type 60 and select seconds from the drop-down list.
5. Configure the Stats function to count the number of non-null source values.
   1. Click the New Aggregations drop-down list, and select count.
   2. Type source in Field/Expression, and num_events_with_source_field in Output Field.
6. Click Validate.
7. Click Start Preview and the Stats function to verify that your data is being aggregated. In this example, we are using a time window of 60 seconds, so your preview data for Stats shows up after 60 seconds have passed between the timestamps of your records.

If your data stream contained the following data:

Following this example, the Stats function would contain the following output:

See also

Functions

Stats

average

count

max

min

sum

etsdc

perc