Hello Isaac,
The CUG rebalance is not based on device count, but rather each device is assigned a score internally. This score is calculated based on the total number of dynamic apps aligned to the device and how many actively monitored interfaces the device has, and the final score is whichever of these 2 values is highest which then becomes the device rating. It then calculates the collector's load, which is the sum of all of the device ratings for the devices aligned to the collector. During a CUG rebalance, SL1 tries to spread out the load evenly based on the device rating and the current collector load. The rebalance will also be affected by whether or not collector failover is enabled in the CUG.
For example, a device with 50 actively monitored interfaces with minimal monitoring may have a higher rating that a Windows server device with full PowerShell monitoring. This is why it may seem that some collectors have more devices than others in the same CUG even after a rebalance.
Antonio Andres
Senior Technical Support Engineer | ScienceLogic