Document creation date:
Document last saved:
The example TimeSpanAnalyzer for the statistics module stores page hits and presents you an average based on timespan settings. The data is stored in blocks of time based on your config settings.
Lets say you view statistics that are three minutes old, if you have a <EPnIntervalCountMinute5> larger than 0, you would get an average value calculated on a five minute interval basis.
The configurable keys in this section are:
In effect, these settings tell your logging mechanism how many blocks of the configured timespan to save. If you, for instance, set <EPnIntervalCountHour> to 48, you can access statistics averaged on a one hour basis stretching back 48 hours in time. If you request statistics older than this, you will get statistics calculated on the next, higher, tier in the calculation tree (in this case it would mean EPnIntervalCountDay). Through time, no data will be actually lost, but the granuality will deteriorate, the precision of the statistics will become less accurate and calculated on an average stretched over larger and larger time spans.
The number you enter for each of these tags tell the logger how many blocks of that specific timespan to save in the database before flushing the data up the pipe to the next, larger, timespan block.
An example configuration could look something like this:
In this sample the logger would store:
|One minute blocks of data||no statistics|
|Five minute blocks of data||two hours back in time (data saved with 5 minute granuality)|
|One hour blocks of data||one day back in time|
|One day blocks of data||approximately two months back in time|
|One month blocks of data||two years back in time|
|One year blocks of data||five years back in time|