Statistics Channel Properties

Top Previous Next

This article cover properties of a single statistical channel.

Source Data

Expression

The Expression property defines what data is the base of statistics. It is evaluated on every change of variable that the channel is built upon.

The expression must evaluate to a number (integer of floating point). If the expression evaluates to any other data type, the system tries to convert it to a number and skips the value if conversion fails. If the expression evaluates to NULL, the system applies NULL Value Handling method to it (see below).

Statistics Channel Expression Resolution Environment:
Default Context	Context that the channel is defined in.
Default Data Table	Current value of the variable that the channel is based on.
Default Row	0 if Use Key Field property is disabled. Number of currently processed row if Use Key Field is enabled.
Environment Variables	Standard variables only.

Key Fields

A channel may include one or more Datasets for a single context variable. Multiple Datasets can be created for tabular variables. In this case each row in the variable is identified by unique string key (Dataset name). See Working With Key section for further explanation.

Values Post-Processing

Statistics channel provides several options for post-processing values returned by channel Expression:

•	NULL Value Processing

•	Out Of Range Values Processing

•	Minimal Value

•	Maximal Value

NULL Value Processing

This setting defines that should the channel do when its Expression was evaluated to NULL. There are two options:

•	Discard. Value will be silently dropped. If no other values were received for a certain period, channel overall value becomes UNDEFINED for this period.

•	Treat As Zero. NULLs values will be converted to zeros.

Out Of Range Value Processing

Specifies how to process the values that are less than Minimum Value or greater than Maximum Value. The following choices are available:

•	Do Nothing. Simply disables minimum/maximum values for the channel.

•	Discard. Drops values that are less than minimum or greater than maximum. If no other values were received for a certain period, channel overall value becomes UNDEFINED for this period.

•	Normalize. Values less than minimum and greater than maximum are converted to the minimum and maximum values respectively.

Minimal Value and Maximal Value

These numbers define discarding/normalization thresholds, as specified by Out Of Range Value Processing option.

When enabled for the counting-type channels (Counter, Derive and Absolute), these numbers define minimal and maximal per-second rates.

Statistics Storage

Basically, statistical data may be stored in files and memory. File-based storage is persistent, while data of memory-based channels is discarded when server is restarted.

Storage

Storage property defines statistics storage type. There are three options:

•	Standard file (Lower memory usage, slower)

•	Mapped file (Higher memory usage, faster)

•

Memory

Memory based storage guarantees maximum performance, but it doesn't preserve data between server restarts and, therefore, has very limited uses.

Here are some guidelines for making a choice between standard files and memory-mapped files:

•	Memory-mapped file caches channel data in memory (borrowed not from JVM, but directly from the underlying operating system), while regular file's data is kept on the disk.

•	Both file-based storage's performance dramatically depend to the operating system, so in most cases it's necessary to perform some tests to finally select one of them.

Channel Type

The channel Type setting defines how are subsequent source values processed to form Primary Data Points for the channel. See Channel Types for details.

Data Aggregation

Statistics module use Aggregation Functions (AF) to convert Primary Data Points into Consolidated Data Points (see RRD Technology for details). There are siz aggregation functions that may be individually enabled/disabled for a channel:

•

Average

•

Minimum

•

Maximum

•

Total

•

First

•

Last

Data aggregation functions are applied to all Archives enabled for the channel. For example, if Average aggregation function is enabled, the system keeps Minutely, Hourly and other averages.

Average

This function calculates an average of Primary Data Points to form a Consolidated Data Point. This is an average of values themselves (for Gauge channel) or value change rates (for Counter and other channel types).

Minimum

This function takes finds minimal Primary Data Point to use as a Consolidated Data Point. This is minimal value itself (for Gauge channel) or minimal value change rate (for Counter and other channel types).

Maximum

Similar to Minimum.

Total

This function sums all Primary Data Points to form a Consolidated Data Point. This is a sum of values themselves (for Gauge channel) or value change rates (for Counter and other channel types).

First

This function takes first Primary Data Point as the value for a Consolidated Data Point.

Last

This function takes last Primary Data Point as the value for a Consolidated Data Point.