Default Retention Policy
By default, the appliance retains per-second data for 7 days, per-minute data for 14 days, and per-hour data for 90 days. However, it is strongly recommended that you specify a data retention policy that suits your business needs. Retention policies are especially important if you plan to retain large amounts of historical data for an extended period. The maximum retention period is two years. Software updates to OS8.6.0 or later will cap any previous retention policy settings to a maximum of two years. Modifications to the retention policy are recorded in the audit log. This default retention policy is effective with software version OS8.6.0 and later.
Enabling a Retention Policy
A retention policy limits the minimum amount of data collected by a granularity of per-second, per-minute, or per-hour over a period of time, or retention period. You could set one retention policy per granularity. For example, you could define a retention policy to save a minimum of one day of data at the per-second interval, a second policy to save a minimum of one week of data at the per-minute interval, and a third policy to save a minimum of one month of data at the per-hour interval. It is recommended that you only keep the minimum amount of data according to your business requirements, including compliance needs.
Per-second data is the highest granularity and requires more memory and disk space than per-minute or per-hour data. Likewise, setting a longer retention period corresponds to storing more data. Monitor the size of the datasets in the BUI by navigating to Analytics > Datasets, and in the CLI by using the analytics datasets context. Adjust retention policies accordingly to meet business requirements while occupying the least amount of space. Retention policies apply to all active datasets; suspended datasets are not affected.
Note that you must increase the retention time with each increase in granularity. For example, you cannot define a retention period of weeks for per-second data and then define a retention period of days for per-minute data.
When you enable a data retention policy, you should assume that old data is immediately removed. For example, if you set a per-second policy for at least three hours, you should assume that any data over three hours old is deleted. In actuality, the appliance periodically deletes old data and may delay removing old data to avoid causing a performance impact. You can significantly reduce the space used by Analytics by setting a retention policy that periodically discards the highest data granularity.
To enable a retention policy, you must have Super-User privileges or have configured authorization within the Dataset scope.
Viewing Retained Data
Worksheet graphs are displayed in the highest data granularity available to the appliance. For example, if your retention policies do not collect per-second data but do collect per-minute data, the graphs are rendered using per-minute data.
See Configuring Users in Oracle ZFS Storage Appliance Administration Guide, Release OS8.7.x for more information on defining authorization scopes for users.
To learn more about the properties for analytics policies, see Data Retention Properties.