Running Hashicorp Nomad on AWS with up to 85% savings
Nomad is a distributed, highly available, data center-aware cluster manager and scheduler that helps deploy applications on any infrastructure, at any scale, on premises or in the cloud. It supports virtualized, containerized, or standalone applications running on all major operating systems, and handles a broad range of workloads.
Recently AWS provided a quick start guide with CloudFormation templates to setup the entire stack on AWS. Running the CloudFormation template by following the instructions documented here creates multiple AWS Auto scaling Groups.
All of the instances in an Auto Scaling group run as On-demand instances.
Cmpute.io is a SaaS solution that balances AWS workloads to achieve On-demand availability at spot prices. These Auto Scaling groups can be migrated to Cmpute.io platform to bring in significant savings.
Cmpute.io can bring up to 85% savings on compute (EC2) costs.
cmpute.io provides a 3 level defense mechanism to ensure that you get deep cost savings without compromising on your application availability / uptime.
- First, Cmpute.io uses historical data and prediction modeling to identify the right bidding strategy and the right instance type in a spot market so that the instances that are launched are not terminated anytime soon.
- Second, Cmpute.io has a continuous learning algorithm that tracks the spot instances that were launched and pre-empts termination by proactively swapping an instance out if the chances of its failure is high.
- Third, Cmpute.io lets users define a minimum On-demand threshold that will ensure that a part of the instances are always using On-demand instances.
These multiple lines of defense model of Cmpute.io ensures even web applications can be run on Spot Instances while realizing large cost savings.
In this post, I will provide details on managing the ‘Nomad Servers’ Auto Scaling group using cmpute.io and bringing in deep savings in EC2 costs.
cmpute.io – Process
Login to your Cmpute.io account and choose the Auto scaling group app in Cmpute.io
In the Auto Scaling group job creation screen, choose the AWS account and select the Auto Scaling group. If you want to run a few of the instances as On-demand instances you can set a Min On-demand Instance count in cmpute.io.
Step 1:- Create a Job in cmpute.io
Step 2:– Review the newly created job details and then click on ‘RUN NOW’
cmpute.io would takeover the complete orchestration of the Auto Scaling group with the Nomad Server from here on.
Step 3:- Complete details of the run can be monitored from the ‘Job summary’ screen. Navigate to jobs and select the job you just started, you will be able to monitor all the details related to instances, savings and instance distribution across zones.
One of the servers was launched as an On-demand instance as specified in this case and then the other four instances as spot instances. The savings would go up to 85% when we add more instances to the cluster.
Similarly other Auto Scaling groups can be migrated to cmpute.io to achieve significant savings. You can start your free trial from here.
Latest posts by Anjali Panwar (see all)
- Running Hashicorp Nomad on AWS with up to 85% savings - June 2, 2017
- How to run Large Transcoding Jobs on AWS using Batchly - July 26, 2016