PERFORMANCE

Understanding how Activity Trends collects data
Activity Logging collects data from the log file (LOG.NSF) and the Catalog task and stores it in the Activity Trends database (ACTIVITY.NSF). The Activity Trends Collector task processes this data and produces the trended data that is used in charting and resource balancing.

The "Trends cardinal interval," "Observation time bucket," and "Proxy data" settings affect Activity Trends.

Trends Cardinal Interval

Trend statistics are based on data gathered during an observation period, which is a 24-hour period from midnight to midnight. Each trend statistic is a weighted running average, which is computed by adding data from a new observation to the existing "trend," or running average, with an exponential weighting.

Consequently, the newest observations are weighted most heavily, and older observations are weighted exponentially less and less in the new computed trend. Keep in mind that increasing the cardinal interval increases the number of recent observations that are heavily weighted, and decreasing the cardinal interval decreases the number.

Observation Time Bucket

Activity Trends stores data in a "time bucket," or array, that represents a distribution of activity across one observation period. When you set up Activity Trends, you specify the size of each bucket, by specifying the number of seconds that make up one bucket. The specified number must divide evenly into one hour. For example, the default is 300 seconds, or 5 minutes; therefore, there are 288 5-minute buckets in one observation period.

Proxy data

By default, the server from which you are running Activity Trends will find the local Activity Trends database (ACTIVITY.NSF). However, you may replicate Activity Trends databases that contain data you want to access. You use proxy data to include the names of other Activity Trends databases that contain trends data from other servers.

See also