Glympse Reduces their Nomad Cluster Costs by 70% with Spotinst Elastigroup
Glympse is a built-for-mobility SaaS leader that helps companies empower their customers by eliminating the anxiety and uncertainty around product and service deliveries. They pioneered real-time temporary sharing technology and built a powerful platform that leverages that intelligence to unite key marketing, commerce, and service engagement points into a single interactive and easy-to-use customer experience.
Organizations are always looking for ways to enhance their customer experience with innovative features. Companies such as Glympse provide those organizations with an innovative way to enhance their customer interactions with technology that delivers shared real-time GPS information for commerce and services. When Glympse services are used in customer-facing applications or websites, customers can instantly know where their service technician, airport shuttle, or grocery delivery driver is without having to contact customer service for updates. When a service such as Glympse is offered to potentially millions of users, costs and scalability will become a concern.
Glympse services currently receive over 1400 requests per second during peak times and that load is spread across multiple Amazon Web Services (AWS) regions. They use Nomad by HashiCorp to manage and deploy their compute infrastructure on AWS. As Glympse grew their customer base, they began to realize the high costs of using public cloud services and the added difficulty of managing a large infrastructure at scale.
To succeed in making their customers happy and grow, Glympse was in need of a solution that could reduce their cloud computing costs, work with their existing Nomad setup, manage the underlying infrastructure, and autoscale their workloads to meet load requirements for peak times. Glympse began looking for solutions to assist with their challenge and began to evaluate Elastigroup.
In order to reduce costs, Glympse previously tried using Spot instances with Spot Fleet and open source solutions, but they found the tools to be not easy to use and required additional management and overhead. When Glympse learned about Elastigroup’s support for Spot instances with Nomad, they decided to try it.
Elastigroup integrated seamlessly with Glympse’s Nomad client/server deployment configuration. Each instance within the Nomad Elastigroup automatically registers itself to the cluster as a client node and will be managed by Nomad and monitored by Elastigroup. Nomad servers are managed independently using your preferred instance type.
Prior to using Elastigroup, Glympse would have to manually scale up their compute infrastructure ahead of peak times to ensure service availability. When peak times end, additional manual intervention was needed to reduce unused compute resources to reduce costs. Glympse found that they would often over-provision compute resources to have a safety net but that led to an increase in unnecessary costs and management. Glympse found a better way to scale their infrastructure up or down with Elastigroup.
Elastigroup provides automatic scaling based on Nomad Job Queue requirements for CPU & Memory to meet the changing demand of services provided through Nomad with the most responsive and cost-effective infrastructure. The Nomad Autoscaler also monitors compute usage to automatically terminate unneeded or underutilized nodes. If a node is not needed for an extended period and is not highly utilized, the Jobs will be scheduled elsewhere and the node will be terminated to defragment the cluster.
Glympse also purchased many reserved instances on AWS in attempts to control costs for an extended period of time. After Glympse began using Elastigroup to manage their infrastructure, it found a couple dozen unutilized reservations. If a reserved instance is not being used, that is money wasted. Elastigroup was able to optimize their reserved instances because it searches for all unutilized Reservations and prioritizes Reservation usage prior to launching Spot Instances, making sure workloads are running on the most cost-effective mix of instances.
“Elastigroup was the easiest solution to use and provided us with the best way to reduce our compute costs while taking a hands-off approach to managing our infrastructure. The Nomad integration and auto-scaling were keys for our success. We look forward to expanding our use of Elastigroup for more of our production and development needs,” John Gruender, VP Engineering, Glympse.
Glympse is now running over 102 Spot and 20 reserved instances managed by Elastigroup, and it only took them two weeks to get up and running. After Elastigroup was used in production, they were able to reduce their cloud computing costs by 70%!
Elastigroup’s Nomad autoscaler provided Glympse with a hands-off approach, less infrastructure management, and simpler operations. In the near future, Glympse plans on using Elastigroup more in production to manage their reserved instances and move their existing Elasticsearch and Kafka workloads into Elastigroup.