Memory Machine™ for Kubernetes

Datasheet

Hot Restart Non-Fault Tolerant Stateful Apps after Pod Kills and Evictions

Kubernetes (K8s) is the de facto cloud operating system for enterprises implementing hybrid and multi-cloud architectures with greater efficiencies. As a result, an increasing percentage of the applications are stateful. However, K8s’ high frequency of pod kills and evictions makes it challenging to operate non-fault-tolerant stateful applications that were never designed for frequent cold restarts. This results in lower user productivity, lower resource utilization, and higher administrative costs. The Memory Machine for Kubernetes Operator is designed to address these type of operational challenges by enabling stateful applications to gracefully resume from the point where they were interrupted.

Key Features

  • Certified Kubernetes operator and sidecar architecture for simple deployment
  • No changes to applications, workflows, or Kubernetes distribution
  • Machine and network socket state, pod and application state, and associated temporary files transparently captured and restored from old to new pod
  • Supports RedHat OpenShift, ROSA, AWS EKS, and other distros
  • Node to node transfer of pods supported within same class of machines
  • Supports enterprise customer subscriptions

Key Benefits

  • Simplifies performance and scheduling of node maintenance operations involving stateful apps
  • Provides higher application availability for stateful applications
  • Save compute costs on the public cloud
  • Increase productivity of users of stateful applications by preserving previous work and eliminating cold restarts/downtime from interrupted operations
  • Improve resource utilization and more sustainable computing by reducing restart and downtime intervals
  • Free up developer resources from re-factoring legacy stateful apps to focus on creating more business value

Use Case 1

Node draining for maintenance

K8s will periodically drain nodes and transfer running stateful apps to other nodes in order apply software patches and perform other maintenance. Memory Machine will automatically checkpoint the node and gracefully migrate the running application state to the new node without loss of previous work or the need for a cold restart.

Use Case 2

Public cloud Cost Optimization Using Spot Instances

Run K8s nodes with stateful apps on Spot Instances to save up to 70% while automatically hot restarting any pre-empted pods on alternative instances.

Use Case 3

Horizontal Scaling

Use the MMK Operator in conjunction with horizontal scaling for automated bin-packing while maintaining business continuity of stateful applications.

Use Case 4

Vertical Scaling

Avoid OOM kills and overprovisioning by automatically resizing pod CPU & memory requests based on application usage. Seamlessly migrate a pod to another node when there are not enough resources on the original node.