Using the statistical model
The statistical load balancing model uses load values that you specified to determine a target datastore.
A load is the weighting factor that is assigned to a datastore for load balancing. The statistical model allows you to assign a load (0 through 100) to each datastore to specify how the workload is distributed among the active datastores. These loads take the place of WLM.
When you assign a weighted value to a datastore, an algorithm converts the weight (1 through 100) to a goal percentage. The default is 1. Only active datastores are considered when setting goal percentages. For information about setting load balancing weights for datastores, see Defining-load-balancing-weights.
- With the statistical method, datastores are considered active when goal percentages are calculated.
- With the WLM method, messages are sent after the first specified cycle time. This happens because the datastore status is undetermined at that time.
The following examples show how the goal percentage is calculated in a 2-minute cycle with datastores becoming inactive and active within a cycle. The Affinity Manager is inactive.
Related topic