Azul – a computer software company that develops runtimes for Java-based applications – had a pressing need for a quicker solution when autoscaling their Kubernetes clusters. Their infrastructure on AWS was unable to properly manage their load spikes, resulting in autoscaling taking a matter of minutes rather than seconds.
CloudWerx was contracted to migrate Azul’s existing AWS workloads to GCP and provide a solution that could optimize performance and auto-scale significantly faster than their previous setup.
The Customer: Azul
Azul is the only company 100% focused on Java, delivering the most trusted Java platform to the modern cloud enterprise. They provide the world’s best commercial support for OpenJDK to their customers by prioritizing their success, maintaining an unwavering commitment to innovation and excellence, and advancing Java through community leadership.
The Challenge: Lightning Fast Autoscaling
Azul stated the goal of achieving “lightning fast” autoscaling for their Kubernetes clusters. They had an incredibly spiky server load profile that achieved desired results in a matter of several minutes as opposed to the preferable several seconds. The customer needed a quicker solution in GCP that was able to handle their load spikes.
The Solution: Migrating to GCP
Azul’s existing infrastructure is on AWS. With a need for a better solution in GCP, CloudWerx was able to create a GKE Standard cluster for the customer to benchmark the load testing on GKE. The goal was to get similar or better performance as EKS. We met the goal by choosing the compute-optimized machines on the GKE node pool.
As a result of replicating their existing AWS infrastructure on GCP, Azul was able to have a secondary environment in GCP with the same, if not better, performance in GCP. The customer was thrilled to have a safe, secure, and fast environment now set up in GCP that they can trust to meet all performance requirements. Ultimately, CloudWerx was able to provide insight and hands-on keyboard work to deploy a better auto-scaling solution on GCP than AWS.