The Serverless Kubernetes Engine
Spotinst Ocean is a Serverless Kubernetes Engine that abstracts the compute layer in a Kubernetes cluster. With Ocean, you no longer have to manage or scale a cluster of virtual machines. Ocean can save you 80% on your compute costs by running on cloud excess capacity.
Managing infrastructure is not a simple task and requires many human resources and skills to do it right. Adding a Kubernetes cluster to manage will add complexity and require even more resources. Ocean’s technology simplifies IT operations and reduces costs. Ocean leverages reliably cloud excess capacity (Spot Instances) to save 80% of the compute costs.
How it works
Ocean integrates with existing Kubernetes control planes or “Kubernetes as a Service” solutions such as Amazon EKS, Google Kubernetes Engine (GKE), Azure Kubernetes Service(AKS) or even vanilla Kubernetes clusters via KOPS and provides an engine for provisioning, managing and scaling the Kubernetes nodes.
Pod Driven Autoscaling
Your VMs size is determined by your Pods & Deployments parameters. Elastigroup re-schedules Pods to optimize your cluster for performance and costs. Read More
Heterogeneous Instance Fleets
Elastigroup utilizes multiple Instance Types from different Instance families and sizes to increase the cluster’s performance and efficiency (including GPU & FPGA support).
Pay 80% less on compute infrastructure
Elastigroup reliably utilizes Spot Instances based on Machine Learning models to predict Spot Instances interruption to maintain high-availability.
Elastigroup uses Spot Instances to reduce costs, however, when Spot Instances are not available, it automatically falls back to On-Demand Instances. When Elastigroup uses On-Demand, it will opportunistically revert back to Spot Instances when capacity is available.
Prioritizes existing Reservations (RIs)
Elastigroup discovers unutilized Reservations in real-time and prioritizes Reservations usage prior to launching Spot Instances.
Enhanced monitoring & auto-healing capabilities
Advanced dashboards to track the Kubernetes cluster utilization and efficiency.