Container adoption is growing, and Kubernetes is becoming the de facto standard of container management platforms. Whether container adoption occurs on-premises, in public clouds, or both, the operational overhead is enormous. IT administrators cannot foresee computing resource demands of applications, so they must reserve more computing resources for a workload than needed. Managing computing resources and optimizing costs on multiple clouds are daunting tasks. Federator.ai, ProphetStor’s Artificial Intelligence for IT Operations (AIOps) platform, provides intelligence to orchestrate container resources on top of VMs (virtual machines) or bare metal, allowing users to operate applications without the need to manage the underlying computing resources.
After Federator.ai is deployed in an OpenShift environment, it learns application resource usage patterns and predicts the needed resources on a per namespace level down to the container level. Federator.ai also provides a dashboard that displays the per-application workload and resource recommendations. Federator.ai for OpenShift delivers the following key features:
Federator.ai applies multiple analytics tools, such as machine learning and signal processing, to predict containerized application and node resource usage as the basis for pod resource recommendations. Federator.ai supports both physical and virtual CPUs and memories.
The application resource demand determines the number and size of pods. Federator.ai utilizes resource usage prediction based on workload patterns to recommend the right pod sizes.
Federator.ai plans cluster-wide CPU and memory allocation for different types of applications according to the policy specified by users.
Federator.ai is designed to work with any OpenShift-operated environment. Federator.ai provides application lifecycle management based on the Operator Framework and works seamlessly with Red Hat OpenShift.
Installing Federator.ai is easy as it works as an Operator on OpenShift.
Federator.ai continuously generates recommendations and learns better with more accumulated metrics data.
Over-provisioned computing resources and the deployment of the incorrect number and/or size of VMs and/or pods are two common issues in a cloud-native environment. Federator.ai addresses these problems by orchestrating resources in multi-cloud environments. As shown in the Figure, Federator.ai optimizes costs for both Day-1 deployment and Day-2 operations. It utilizes metrics stored on Prometheus, collected by OpenShift, to predict resource consumption dynamically and recommends the right amount of resources for pods, providing a 20 – 70% reduction of wasted resources for a typical workload, as well as preventing under-provisioning of resources for mission-critical workloads. Users can stack up the predicted pod resources to determine the right number and size of VMs to deploy and enable the automatic execution of these recommendations.
With Federator.ai, users no longer need to specify the CPU and memory requests and limits for each container. It recommends optimal pod configurations. The direct effect is that the configured resources will accurately and dynamically match the workload. It also effectively reduces occurrences of under-provisioned issues, such as out-of-memory (OOM).