Elastigroup & Google Getting Started: Create A Cost-Efficient Auto Scaling On GCP

As we previously announced our integration with GCP, We are happy to share a getting started post that describes the flow of creating a cluster using Spotinst on Google Cloud Platform.

Do more with Google Cloud.
Save 70% of your infrastructure by running cost-aware clusters.

In this post, we will demonstrate how to create a cost-efficient Auto-Scaling in GCP using Spotinst Elastigroup.

This post assumes that you already have a Spotinst account, and that you have connected your Google Cloud account with Spotinst.

Elastigroup is a perfect fit for Autoscaling clusters and environments that run behind HTTP/Network Load Balancers

First thing first,

Choose your environment. Whether if its an autoscaling or HTTP Load balancer cluster.
Then login into the Spotinst Console and go to “Elastigroup” screen and press “Create”.

Select name and description for your cluster

Capacity Configurations

Maximum, Minimum and desired(target) number of instances

Blended Clusters of Preemptible and Regular VMs

You can simply control the percentage of Preeamptbile VMs vs Regular VMs in the same cluster

Compute Configurations

Select your Network, desired Instance sizes and types (Standard, High-Memory, High-CPU) , Zones(Google Data Centers), Operation System disk and of course your HTTP Load Balancer Backend Service.

Startup Script, Tags and Meta Data

Add tags and metadata information that will be applied into the instances in the cluster

Scaling Policies

Select your scaling measurements, either CPU Utilization, Network Bandwidth, or any other Metric that Google Cloud Monitoring exposes.

Once done, continue to the review step and hit the “Create” button.

Starting from now you will be running an Elastigroup, that provisions GCE VMs within your Google secured account.
Those instances are auto-scaled, auto-healed and auto-recovered automatically by Spotinst.

Preemptible VMs Interruptions Managed By Elastigroup

Based on a prediction algorithm and in order to minimize and reduce the Preemption rate of Google cloud, Spotinst automatically spins new machines before failures happen or before regular termination (24h) of each Preemptible VM to ensure a smooth and steady workload.

Best Regards,

The Spotinst Team.