What specific metrics should we use to scale our healthy edge microgateway instances in and out due to varying load?

We are trying to determine what metrics it makes sense to scale our edgemicro instances on. Right now we can chose some prededfined AWS metric, like avg CPUUtilizaiton. Would you recommend scaling on a different metric, like for example, any of the metrics returned from the HealthCheck plugin?

(We are asking about scaling the desired number of healthy instances in and out due to load - as opposed to asking about how to achieve a particular desired capacity with disastor recovery)

0 3 319
3 REPLIES 3

Former Community Member
Not applicable

I think the healthcheck plugin is a good alternative. It provides the health of the node MG is running on.

Thanks a lot for the response! @srinandans


I think I could be more clear in my question, I'll edit it.

We are asking about the appropriate metric to use to scale a certain desired capacity of (already healthy) instances in and out to handle variations of network traffic. (aka, What should I measure to appropriately change this desired capacity?)
Example metric: Average CPUUtilization.

I think the metrics returned by the healthcheck plugin is memory at the node process level instead of machine CPU numbers. (Average CPUUtilization wouldn't necessarily translate to telling us if the application is performing near the upper limit of what load it can handle)


(We are not asking about what to use to see if an instance is healthy so that we can apply automated disaster recovery to an instance.) (aka, we are not asking How do I know if an instance is healthy?)
(Example: the edgemicro healthcheck plugin)

Regards,



Former Community Member
Not applicable

I think the healthcheck plugin is a good alternative. It provides the health of the node MG is running on.