Autopilot configuration: Key concepts

Four thresholds apply to image caches:

Day - Amber
Day - Red
Night - Amber
Night - Red

Autopilot estimates the average incoming volume, then uses the user-configurable thresholds (in hours) to calculate how many MB of free cache space that translates to. Autopilot tries to free up enough cache to bring the system back to the Amber or Green zone.

Each site must estimate their average volume and how many hours worth of free cache space they want Autopilot to maintain. The latter gives them a sense of how long the cache can last if archive and Autopilot fail before the cache fills up. With the total size of cache available, the average volume gives them a sense of approximately how many days of studies that cache can hold.

The configuration itself is relatively simple. The difficult part is understanding the activities at the site and estimating the average volume.

How are the thresholds determined?

The thresholds are determined by a combination of:

User-defined thresholds (in hours)
Effective study storage rate (in MB/hour)

How is the effective storage rate determined?

Using SQLPLUS, query the mvf_ap_statistics table for the following parameters:

effective_study_arrival_rate (studies/hour)
objects_per_study (objects/study)
storage_units_per_object (KB/object

Effective study storage rate is expressed in KB or MB per hour.

From the previous section:

Free cache space required (MB) = Z hours (user-defined threshold) * Y MB/hour
Threshold = Total cache space – Free space required

How many studies will Autopilot delete?

Autopilot will delete as many studies as necessary to bring the cache back to Amber or minimum percentage occupancy.

Let X = Average size of a study (MB) and Y = Free cache space required (MB)

Number of studies to delete = Y / X