Antrea: Kubernetes Networking and Security on an Open vSwitch Data Plane
Antrea is an open-source Kubernetes networking and security project built on Open vSwitch, providing pod networking, policy enforcement, observability integration, and gateway controls for production clusters.
At a glance
- What is it?
- Antrea is a Kubernetes-native networking project that uses Open vSwitch for pod networking, policy enforcement, and observability. This review covers its architecture, deployment, limitations, and alternatives for engineers evaluating it.
- Who is it for?
- Adopt Antrea if you need a Kubernetes-native CNI that leverages Open vSwitch for high-performance policy enforcement and supports Windows nodes or multi-cluster setups. Avoid it if your cluster cannot load the OVS kernel module or if you prefer a simpler, iptables-based CNI.
- Can I use it commercially?
- Yes. Apache-2.0 is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository received new commits within the last day.
- What is it written in?
- Mainly Go, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 15, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
The Problem Antrea Solves
Antrea addresses the need for a Kubernetes networking solution that is both Kubernetes-native and high-performance. Standard Kubernetes Network Policies rely on iptables or similar mechanisms, which can become inefficient at scale. Antrea replaces the data plane with Open vSwitch, a programmable virtual switch, to handle pod networking, service load-balancing, and policy enforcement. The project targets production clusters that require efficient policy enforcement, hardware offloading for demanding workloads, and support for both Linux and Windows nodes. It is for teams that want a single CNI that does not compromise on performance or feature breadth.
How Antrea Works: Open vSwitch as the Data Plane
Antrea operates at Layer 3/4 and uses Open vSwitch as the networking data plane. The control plane extends Kubernetes APIs to manage the OVS data plane on each node. According to the README, Antrea leverages OVS to implement Kubernetes Network Policies efficiently and to handle Service load-balancing. The architecture is Kubernetes-native, meaning it uses Kubernetes libraries and APIs for its own implementation. OVS also enables hardware offloading, which is a key differentiator for workloads that need line-rate processing. The data plane is consistent across Linux and Windows nodes, thanks to OVS's portability, which simplifies operations in mixed clusters.
Getting Started: Deployment and Prerequisites
Deployment is described as simple: apply a single YAML manifest. The README points to the getting-started document for details. Before that, you must meet two prerequisites. First, the Kubernetes cluster must run version 1.23 or later, and NodeIPAMController must be enabled. For kubeadm clusters, this means specifying --pod-network-cidr. Alternatively, you can enable and configure the NodeIPAM feature of Antrea Controller. Second, the Open vSwitch kernel module must be present on every node. This is a hard requirement; without it, Antrea cannot function. The single-manifest deployment is attractive, but the OVS kernel module prerequisite means you cannot run Antrea on nodes where that module is unavailable or unloadable.
Policy Model: Beyond Kubernetes Network Policies
Antrea's policy model is a major feature. It builds on Kubernetes Network Policies but adds policy tiering, rule priorities, cluster-level policies, and Node policies. These are collectively called Antrea Network Policies. The tiering and priority features give administrators finer control over how policies are applied and which rules take precedence. Cluster-level policies apply across namespaces, and Node policies can enforce rules on non-Kubernetes nodes, including VMs and bare-metal servers. This is a genuine extension of the standard policy API. However, the added complexity means you must learn a new policy model beyond the basic Kubernetes NetworkPolicy resource. The README does not detail the exact API resources, so you would need to consult the documentation for specifics.
Observability and Multi-Cluster Features
Antrea includes troubleshooting and monitoring tools: a CLI and UI for packet tracing, policy analysis, and flow inspection. It exposes Prometheus metrics and supports exporting network flow information to collectors. The README mentions Theia, a separate project, for visualization in Grafana dashboards and policy recommendations. For multi-cluster environments, Antrea offers federation of multiple clusters with a unified data plane and security posture, including multi-cluster Services. These features position Antrea as more than a basic CNI; it is a platform for network observability and cross-cluster operations. The trade-off is that these capabilities are spread across multiple projects (Antrea, Theia, Nephe), so evaluating the full stack requires looking beyond the main repository.
Encryption and Platform Support
Antrea can encrypt inter-Node Pod traffic using IPsec or WireGuard tunnels. This is an optional feature, and the README does not specify default behavior. The choice between IPsec and WireGuard involves trade-offs in performance, compatibility, and key management. Antrea also supports Windows nodes, which is rare among CNIs. The same OVS data plane works on both Linux and Windows, reducing the need for separate implementations. However, Windows support may have feature parity gaps; the README does not clarify which features are available on Windows. The encryption and Windows support are significant for enterprises with mixed OS clusters or strict security requirements.
Limitations and When Antrea Is the Wrong Tool
The most obvious limitation is the dependency on the Open vSwitch kernel module. If your nodes run a kernel without OVS support, or if your security policy forbids kernel modules, Antrea is not viable. The README also requires NodeIPAMController to be enabled or the Antrea Controller's NodeIPAM to be configured, which adds a setup step. The complexity of the policy model and the multi-project ecosystem (Theia, Nephe) can be overwhelming for small clusters that only need basic networking. For a simple cluster with a few nodes, a lighter CNI like Flannel or Calico (in iptables mode) might be easier to operate. Antrea is designed for production clusters with demanding workloads, not for minimal setups.
Alternatives and Comparison
The main alternative is Calico, which also provides networking and policy but uses eBPF or iptables as its data plane rather than Open vSwitch. Calico is more widely adopted and may have a simpler deployment model, but it does not offer the same hardware offloading capabilities that OVS provides. Another alternative is Cilium, which uses eBPF for high-performance networking and policy enforcement. Cilium is known for its L7 policy and observability features, but it requires a newer kernel and does not support Windows nodes as maturely as Antrea claims. The key difference is the data plane: Antrea's OVS is a user-space switch with kernel module, while Cilium and Calico's eBPF are in-kernel. This affects performance, feature set, and operational constraints.
Maintenance and License Considerations
Antrea is licensed under Apache-2.0, which is permissive and allows commercial use with attribution. The project is actively maintained, with recent releases v2.7.0, v2.6.3, and v2.5.3, all pushed on the same day, indicating a steady release cadence. The repository is not archived. Maintenance cost for adopters includes keeping up with releases, ensuring OVS kernel module compatibility with kernel updates, and managing the additional components like Theia if you use observability features. The README does not provide upgrade instructions, but the changelog directory suggests that release notes are available. The Apache-2.0 license imposes no copyleft obligations, but you should review the license text for any patent clauses, as is typical.
Editorial conclusion
Adopt Antrea if you need a Kubernetes-native CNI that leverages Open vSwitch for high-performance policy enforcement and supports Windows nodes or multi-cluster setups. Avoid it if your cluster cannot load the OVS kernel module or if you prefer a simpler, iptables-based CNI. Before adopting, verify that NodeIPAMController is enabled or configure Antrea Controller's NodeIPAM, confirm OVS kernel module availability on every node, and test the encryption mode (IPsec or WireGuard) that fits your security requirements.
Community notes