Setting min, max parameters for Horizontal VM scaling

[Sai Koti Reddy Danda]

Hello,

 

I am planning to use horizontal VM scaling functionality from CloudSim Plus. I have the following doubts :

 

1. Is there a way to set and update the minimum and maximum parameters for horizontal VM scaling in CloudSim plus ? 

 

These parameters are to set the boundaries for Autoscaling. Maximum limit lets Auto Scaling scale out the number of instances as needed to handle an increase in demand. However, It can not add more than the maximum number of VMs in any case. The minimum limit helps ensure that you always have a certain number of instances running at all times.

 

2. Are there any other metrics supported by CloudSim Plus apart from CPU, Memory and BW. For example, the application metrics like latency etc.. that are running on VMs. If not, is there a way to incorporate them ?

 

Can anyone please kindly clarify my doubts ?

 

Thanks

Sai Koti Reddy Danda

[Prof. Manoel Campos]

Hello Sai,

 

I am planning to use horizontal VM scaling functionality from CloudSim Plus. I have the following doubts :

 

1. Is there a way to set and update the minimum and maximum parameters for horizontal VM scaling in CloudSim plus ? 

The scaling classes in CloudSim Plus provides a base for you to implement any scaling algorithms you want. The classes are completely generic so that you can implement that in your own way. They just provide the basic functionalities to track VM under and over utilization.

Check LoadBalancerByHorizontalVmScalingExample. The createHorizontalVmScaling() method sets an overload predicate which is automatically called to check if the VM is overloaded (in order to create new ones for load balancing). There you can define your own logic, including setting min and max number of VMs to create. It's up to you the way you will implement that.

  

2. Are there any other metrics supported by CloudSim Plus apart from CPU, Memory and BW. For example, the application metrics like latency etc.. that are running on VMs. If not, is there a way to incorporate them ?

Regarding the horizontal VM scaling, you can use any resource and/or metric  to define if a VM is overloaded or not. The method isVmOverloaded() in the LoadBalancerByHorizontalVmScalingExample is a simple example of how you define a VM overloaded, considering just the CPU utilization. You can add any other resource or metric you want there.