SAK base_capacity и above_base_capacity - Evanto/qna GitHub Wiki

base_capacity vs above_base_capacity

On_demand_base_capacity - is a minimum number of on-demand instances in a cluster. (set in numbers)

On_demand_percentage_above_base_capacity - is a percent of on-demand instances to use in a cluster after the on-demand base capacity has been met (was reached). (set in percent). When 0, it’s automatically reached.

Example:

I you set:

base_capacity = 5
above_base_capacity = 10

Then:

  • 5 is a number to aim for at the moment of cluster launch. Once it is launched, above_base_capacity dictates your desired percent of on-demand instances in a cluster that you want to maintain.

  • This means that after base_capacity is reached (you have 5 Commons in a cluster), 10% of instances will be on-demand and 99% will be spot.

  • The operator node is always on-demand by default.

ASG group capacity parameters:

  • Minimum count of instances that might be created inside of the ASG (in an EKS cluster) (required)
  • Maximum count of instances that might be created inside of the ASG (in an EKS cluster) (required)
  • Your desired capacity (optional)
  • On-demand base capacity (how much on-demand instances)
  • On-demand above base capacity

Definitions from Auto Scailing

Desired Capacity [optional]: The desired number of Amazon EC2 instances that should be running in the group. Must be between the minimum and maximum size (greater than or equal to the minimum size and less than or equal to the maximum size specified for the ASG). When specified, that will be the total number of instances launched right after creating ASG. - Сколько всего инстансов я хочу запустить при старте ASG группы (желаемая цифра в пределах мин и макс значений). Мин и макс задают крайние пределы, а тут - желаемая цифра между ними.

Links:

Cortex Gitter

Cortex - Spot Instances Configuration Example

YT: Leveraging Amazon EC2 Spot Best Practices Using the New Capacity Optimized Allocation Strategy

Auto Scailing The Ultimate Guide to Deploying Kubernetes Cluster on AWS EC2 Spot Instances Using Kops and EKS

AWS Docs

Auto Scaling Groups with Multiple Instance Types and Purchase Options

Setting Capacity Limits for Your Auto Scaling Group

How ASG (Auto Scaling Groups) Work

ASG - auto scaling group. ASG is a logical group of instances for your service. Auto scaling group instances have multiple purchase options and instance types. minimum and maximum are bound to the number of instances that can be in ASG.

What ASGs do:

  • simple automatic capacity provision
  • replace unhealthy instances
  • balance capacity across availability zones
  • scale infrastructure up and down
  • automatically launch and terminate instances to meet your desired capacity
  • support multiple purchase options and instance configurations
  • one ASG can combine multiple purchase options and instance types across several availability zones

So in an ASG group you have:

  • Minimum number of instances that can be in an ASG
  • Maximum number of instances that can be in an ASG
  • Your desired capacity
  • On-demand base capacity (how much on-demand instances)

There are several types of Amazon EC2 Autoscaling:

  • Manual Scaling
  • Scheduled Scaling
  • Dynamic Scaling
  • Predictive Scaling