Introduction

What is HDR

Historical data recording (HDR) refers to a group of management features that allow you to collect SBC statistics, save them to a CSV file, and SFTP this file ito one or more designated servers. The statistics are grouped so that you can refer to a set of statistics by simply invoking their group name. For example, the system statistics are in a group called System; interface statistics are in a group called Interface. Each group contains several metrics.

The following table describes the type of HDR statistics that the SBC can collect.

HDR Statistics	Description
Group Name	The name of the group that contains the HDR statistics. This name is similar to the current SBC ACLI parameters. For example, system, interface, session-agent, session-realm, etc. The SBC uses the group name when generating the CSV file (for example, system.csv, interface.csv, etc.).
Group Statistics	Various statistical parameters within a group. These statistical parameters appear in the first record of the header in each CSV file (for example, in the system.csv file, the header would include the statistic headings of CPU Utilization, Memory Utilization, Health Score, etc.).
Type	Type of statistical parameter. This document makes use of the following data types: counter - A counter is an integer with a minimum possible value of 0 and a maximum value of 4294967295. A counter is always increasing in value or remaining unchanged. It decreases only in response to reaching its maximum possible value, at which point it’s next value (when the next counted entity or event occurs) will be 0. gauge - A gauge is an integer with the same bounds as a counter. However, it does not always have an increasing value. Its value may go up or down. period - A period type represents a value determined as the sum of a number of events which occurred during a specified window of time, or stated alternatively, a time period. There are three windows defined, the "current window", the "previous window", and the "sliding window". For more information about these windows, see Windows of Time. config - For the config type, the value for this field comes from a configuration record. timeticks - For the timeticks type, each tick is 1/100th of a second. string - A statistic type pertains to statistics that display as an alpha-numeric character string.
Timer Value (seconds)	For period statistics (statistics that use a period timer), this is the default value, in seconds, of the timer. This value is usually not configurable. However, this value may fall within a range of values if applicable.
Range	The range of values that a group statistic may use when the SBC collects statistics.

For descriptions of specific HDR Statistics and values, see HDR Groups and Group Statistics.

Collection Interval and Push

In your HDR configuration, you can set:

the groups for record collection
the frequency of collection
the frequency of pushes

The number of records in a push equals the push interval divided by the sample interval time multiplied by the number of groups, plus one header per group.

( ( push interval ÷ sample interval time ) x number of groups ) +1 header record per group = number of records in a push

For example, if you set a push interval time of 60 minutes and a sample interval time of 5 minutes, with 10 groups, the SBC would send 120 group records plus 10 header records (for a total of 130 records) for each push:

[(60 ÷ 5) x 10] +10 = 130

You can configure an option parameter (disabled by default) that instructs the SBC to send a trap when data has been successfully pushed. This trap is defined in the ap-smgmt.mib. It contains the name of the node that successfully pushed the HDR file to an SFTP server, a unique file name for the HDR file that was pushed, and the IP address of the push receiver (configured in the global collection configuration). For more information about the HDR SNMP traps, see the product-specific SBC MIB Reference Guide.

Note:

The SBC clears all local records after a successful push, after 3 consecutive push failures, and after a system reboot.

Windows of Time

Many tables in this guide identify data types of counter, gauge, config, timeticks, and period.

A period type represents a value determined as the sum of a number of events which occurred during a specified window of time (or a time period). There are three possible windows in which events can occur:

current window
previous window
sliding window

The current window is the window during which events are currently being accumulated. The current window ends D seconds after the start of the current window, where D is the duration of the window. The current window is always less than D. Once the current window becomes of size D, it becomes the previous window, and a new current window is started.

The previous window is the time period of duration D which ended at the start of the current window.

The sliding window marks the period of time for the previous window, D, plus the time passed in the current window.

For example, if the default timer is set to 100 seconds, the sliding window represents the completed previous window (period 1 shown below) PLUS the time accrued in the current window (period 2 shown below). The sliding window here is 175 seconds. At the end of 199 seconds, the current window will become the previous window and the sliding window will begin counting up from the 100 second mark.

The statistic counter windows diagram is described above.

For the session-agent and session-realm HDR groups, the default timer value is 30 seconds. Thus the current window is between 0 and 30 seconds. When the current window reaches 30 seconds it becomes the previous window and a new current window is initialized. The initial sliding window is 30 seconds, and after that, the sliding window is between 30 and 60 seconds.

If the sustain-rate-window parameter is not configured, the default window size is 30 seconds. If the sustain-rate-window parameter is configured, the default period is set to the value of the sustained-rate-window.