Baseline data provides attribute values during each hours of the day of the week. Creating views for Baseline data is relatively easy for most monitor types as data is organized in individual (horizontal) tables. This enables one to one mapping of a Baseline table to a view for each monitor type that is available in Infrastructure Management system. Baseline dataviews are named as <tablename prefix>_BL_VIEW.
Each of these baseline dataviews contains the following information:
Monitor instance ID. This is a unique number generated for each instance. The value is assigned by Infrastructure Management system during monitor creation. This value is internal and fixed. This number identifies all other data for each instance.
This indicates the hour of the day of the week. BL_TIMESLOT_DESCR_VIEW has textual description of each time slot. This view has two columns: TIMESLOT and DESCRIPTION.
Stats Attribute NameAVG_Stats Attribute NameHIGH_Stats Attribute Name_LOW
This contains baseline values for the above TIMESLOT. Each attribute of this monitor type has three baseline values. See Baselines for details on how baseline values are computed.
Hourly data and rate data
Every hour, the data collected by a monitor is consolidated into a set of data points for that hour. In other words, the high frequency minute-by-minute (raw) data is converted into a statistical data for each hour. This serves two purposes:
1) Reduce the amount of space required for data storage
2) Provide historical data for trend analysis and reports
The converted data is called rate data. The rate data is divided into three categories:
1. High data (Maximum value), calculated as the upper 90th percentile of raw data for each hour
2. Mid data (Average value), calculated as the 50th percentile of raw data for each hour
3. Low data (Minimum value), calculated as the lower 10th percentile of raw data for each hour
When you plot a graph that covers more than 24 hours, it shows the average (mid) rate data in the graph.
Baseline calculation and rate data
The baseline algorithm calculation places a heavier weight on older data than it does on newer data. The baseline data is also divided into high, mid, and low categories. When you plot a graph with the baseline indicator selected, it shows only the high and low baseline in the graph. The High data can be displayed by selecting the Peak indicator and re-plotting the graph. You can only plot one attribute when any indicator is displayed.
Raw data is only retained for eight days by default. Rate data is retained for 90 days by default. You can use the
pw log period command to adjust the retention period for both raw and rate data. When you plot a graph data older than 24 hours, it is a plot of rate data. If raw data retention is increased to 48 hours, the rate data seen in graphs are older than 48 hours.
If the data is spikey in nature (i.e., Poisson distribution), the rate data is calculated as the arithmetic mean. The product refers to this as a normal rate calculation. Log File Monitor and Event Log Monitor are examples of monitors that produce this kind of data.