Configuration
The application.yml file has all properties required to run BIAS Autoscaler on Google Cloud. Bear in mind it needs to run in the same cluster you are performing the autoscaling. It you wish to run it in a different cluster, you need to configure the Google Cloud SDK to authenticate the pod/VM on your cluster.
Properties
autoscaler:
project: # name of your project
zone: # zone of your project
region: # region of your project
machine-type-ondemand: # VM type for the regular instacens
instance-group-ondemand: # instance group for regular instances
monitoring-group-ondemand: # group ID for regular instances
machine-image-ondemand: # name of image for regular instacens
machine-type-burstable: # VM type for the burstable instacens
instance-group-burstable: # instance group for burstable instances
monitoring-group-burstable: # group ID for regular instances
machine-image-burstable: # name of image for burstable instacens
backend-service: # name of the service for load balancing configuration
scaling:
maximum-regular-instances: # max # regular of instances
maximum-burstable-intances: # max # burstable of instances
maximum-instances: # min # of instances
minimum-regular-instances: # min # regular of instances
minimum-burstable-instances: # max # burstable of instances
current-regular-instances: # total # of regular insntaces for the startup
current-burstable-instances: # total # of burstable insntaces for the startup
cpu-utilization-burstable: 0.4 # CPU weight of burstable instances ( from 0 to 1)
probability-queueing: 0.1 # refer to scalling policy SR Rule
requests-samples: 3 # waiting time to new scaling
cpu-samples: 3 # waiting time to new scaling
autoscaler-decision-interval: 60s # frequency for running the autoscaler
autoscaler-scale-waiting-time: 90 # time by witch the autoscaler will wait to the next scale out/in in seconds
mu: 1000 # mu is the capacity in minuted for the SR Rule where R = arrival/mu
m: 1.0 # overprovisioning constant using burstable instances