Storage of dimension values
When values are detected, they may be stored based on the following configuration options, which operate independently of each other.
- These options are selectable in the dimension definition. See 1. Initial configuration for URL (Normalized).
Logging of dimension values
If logging for the dimension is enabled, all observed values for the dimension are stored in the database. Each instance of each value is counted when detected.
Values stored in the dimension logs are removed after two weeks, so the logs can be kept to a manageable size.
These logs can be used to build up whitelists, blacklists, and group lists while keeping down data growth.
Whitelists versus observed values
For the Values to Record
setting,
you can choose one of the following options:
- Option
- Description
Whitelist Only
- When this option is selected, only the detected values that are
matched against the whitelist that you defined and uploaded are allowed
to be recorded for the dimension. All other values are recorded as
an
[others]
value or, if the maximum number of values per hour was reached (Max Values Per Hour
), the[limit]
value.- For more information about these dimension constant values, see Dimension constants.
For the
URL (Normalized)
dimension, the goal is to build a representative whitelist of the URLs of the web application. The remainder of this page works towards building a useful whitelist for this dimension.
- For more information about these dimension constant values, see Dimension constants.
For the
Whitelist + Observed Values
- When this option is selected, values that match whitelisted values
are recorded, as well as any other value detected in the capture stream.
The net effect is that each URL value for each hit is captured by
the
URL (Normalized)
hit attribute, which then populates theURL (Normalized)
dimension. Each of these values is then recorded.Note: For high-volume dimensions theWhitelist + Observed Values
can generate unbounded growth of the dimension tables in the database. This setting can be dangerous to the overall health of the Discover system, as these observed values may not be purged for some time. Particularly for high-volume dimensions, this setting should be avoided. See Data Management for Dimensions.