HYPR Achieve 78% Monthly Savings Using Spot Elastic Beanstalk Integration
HYPR moved from On-Demand to spot instances and save 78% on their compute costs, using Spot’s Beanstalk integration
HYPR’s Challenge: Fast Phase Growth and AWS Elastic Beanstalk Management
HYPR runs a large scale of Elastic Beanstalk environments based on on-demand instances. As a result of a fast phase growth, they needed to scale up their instances at once. They had less than one month to scale, activate and perform ongoing health checks and version releases for all their Beanstalk environments. “It was very challenging”, says Alon Levi, Software Engineer at HYPR, “we had to increase our activity in such a short time and many actions had to be done in order to complete the mission successfully. Managing our servers was complicated and time-consuming, and the cost was extremely high.”
During this process, HYPR decided to look for a solution that both simplifies the management of their Beanstalk infrastructure and dramatically reduces their costs. “We had to make a change, we’ve heard about Spot and decided to give it a try”.
77% Compute Cost Reduction and Ease of Management with Spot
“The transition was easy and fast. With the support of the Spot team, we migrated all of our Beanstalk environments to spot instances using the Elastigroup by Spot platform. We immediately saw a significant cost reduction and experienced a considerable ease of operation and management.”
It took HYPR only a matter of minutes to migrate their production environment using the Spot import wizard. The migration process automatically copied all the Auto-scaling groups’ configuration and created an Elastigroup with the same configuration. It simultaneously removed the Auto-scaling group policies and suspended the
ReplaceUnhealthy processes. This meant that the Beanstalk environment resources remained untouched and HYPR could keep on updating their application in the same way, using
Elastic Beanstalk AWS CLI commands, as Elastigroup keeps track on the changes and rolls new changes automatically when needed.
HYPR shifted the Elastic Beanstalk production environment immediately. “We decided to use the momentum as we’ve heard about Spot. It was a risk but fortunately, I can summarize it as a great success”.
Spot Solution Enabled Efficiency
Following this success, HYPR added another environment of 150 servers to Spot. As a result of the cost reduction, HYPR could allow to scaling up another 100 spot instances.
“In addition, we moved to bigger and more powerful instances than what we used before the transition to Spot.”
What’s Next For HYPR? ECS & EMR
HYPR started to migrate their ECS clusters to Spot once the ECS Auto-scaler had launched. The Auto-scaler gives HYPR a fully managed solution without the need to configure scaling policies and makes sure the scaling and the underlying infrastructure is as efficient as possible for running their
As a next step, the team will use Spot EMR Integration, helping tackle a big spend for HYPR.