Feature Overview

This feature introduces the ability to utilize swap space at the node level within OpenShift clusters running OpenShift Virtualization (CNV). By enabling swap space as a prerequisite on cluster nodes, workloads can achieve higher density and resiliency, particularly in memory-constrained environments.

The feature will allow OpenShift Virtualization to better manage overcommit scenarios, avoid unnecessary workload evictions, and improve workload placement efficiency without requiring additional physical memory resources.

This feature is meant to replace the "WASP" agent with node based SWAP enablement.

Goals

• High-level Goal Statement:
  Provide OpenShift Virtualization users the ability to run more VM workloads per node by leveraging node-level swap, thus improving hardware utilization and reducing memory pressure risks.

• Who Benefits from this Feature, and How?
  • Cluster Operators / Administrators: Gain improved workload placement flexibility and better cost efficiency on existing hardware.
  • Application Teams / Developers: Can run more workloads with reduced risk of eviction during memory spikes.
  • Business Stakeholders: Achieve better return on infrastructure investment with higher density of workloads.
• Current State vs. Future State:
  • Today: OpenShift Virtualization workloads are constrained by physical memory availability on nodes. Memory oversubscription leads to workload scheduling failures or node pressure conditions.
  • With SWAP: Nodes can safely oversubscribe memory, leveraging swap space to handle bursts, allowing higher workload density and more efficient scheduling without killing workloads right away.

Requirements

Questions to Answer

• What are the performance tradeoffs when using swap for virtualized workloads?
• Should this feature be opt-in (admin must explicitly configure) or default when swap is available?
• How would swap be enabled on an already installed node easily?
• How does swap usage interact with QoS guarantees in Kubernetes?
• What observability tools will be provided to track swap usage?

Out of Scope

• Automatic configuration of swap on nodes (swap must be pre-enabled at installation time of the node).
• Non-virtualized workloads explicitly optimized for swap.

Background and Strategic Fit

OpenShift Virtualization enables organizations to consolidate VM and container workloads on the same platform. To improve workload density and efficiency, this feature leverages OS-level swap as a safety buffer for memory oversubscription. This aligns with OpenShift’s strategy to maximize infrastructure utilization, workload consolidation, and hybrid cloud cost efficiency.

Assumptions

• Node swap will be enabled and managed by the cluster admin prior to use.
• Minimal performance degradation from swap usage is acceptable compared to workload eviction.
• Supported and tested primarily for OpenShift Virtualization (KubeVirt) VMs, not general-purpose containerized workloads.