TL;DR — Cilium is the CNCF Graduated CNI plugin that uses eBPF to provide networking, network policy, load balancing, and observability for Kubernetes — all in the kernel, no iptables. For AI workloads it matters because it handles the management-plane networking (pod-to-pod, pod-to-API, ingress) with high performance and fine-grained security, while RDMA/SR-IOV handles the data-plane GPU-to-GPU traffic.
What it is
Cilium is an open-source Kubernetes CNI (Container Network Interface) plugin built on eBPF. It replaces kube-proxy and iptables with eBPF programs that run directly in the Linux kernel, providing networking, L3/L4/L7 security policies, transparent encryption, service mesh, and deep observability (via Hubble). It's a CNCF Graduated project and the default CNI for GKE, EKS, and AKS. In the AI Native landscape it's in AI Native Infra › Network.
Why it exists
Traditional Kubernetes networking uses iptables — which becomes a bottleneck at scale and gives you almost no visibility into what's happening. Cilium moves networking logic into eBPF programs in the kernel: faster packet processing, identity-based security (not just IP-based), and deep per-flow observability without sidecar proxies. On AI clusters, where thousands of pods do parameter syncs and API calls, this efficiency and visibility matters.
Fig 1 — Cilium provides the management-plane networking for all pod traffic via eBPF in the kernel.
How it works
Cilium installs as a DaemonSet. On each node, it attaches eBPF programs to network interfaces and socket hooks. These programs handle routing, NAT, load balancing, and policy enforcement — all without leaving the kernel. Identities are assigned to pods based on labels (not just IPs), and policies reference those identities. Hubble, the observability layer, taps the same eBPF data path to give per-flow visibility.
Key features
- eBPF datapath — replaces iptables and kube-proxy; faster at scale.
- Identity-based security — network policies based on pod labels, namespaces, DNS, not just IPs.
- L7 visibility — HTTP, gRPC, Kafka protocol-aware policies and observability.
- Hubble — real-time network observability, flow logs, service dependency maps.
- Transparent encryption — WireGuard or IPsec between nodes, zero config.
- Gateway API — native support for Kubernetes Gateway API (ingress, egress).
- Multi-cluster — ClusterMesh for cross-cluster pod connectivity and service discovery.
Quick start
Install via Helm on a Kubernetes cluster:
helm repo add cilium https://helm.cilium.io/
helm install cilium cilium/cilium \
--namespace kube-system \
--set kubeProxyReplacement=true \
--set hubble.relay.enabled=true \
--set hubble.ui.enabled=trueAfter install, all pod networking flows through Cilium's eBPF datapath. Use hubble observe to watch live traffic flows.
When to use, when to skip
Use it as the CNI for your AI cluster's management-plane networking — pod-to-pod, pod-to-service, API traffic, ingress. It handles the TCP/IP layer efficiently and gives you security policies and visibility. On GPU clusters, Cilium manages the control/management traffic while RDMA and SR-IOV handle the high-bandwidth GPU-to-GPU data plane.
Skip it only if your cluster already has a CNI you're happy with and don't need the eBPF benefits. Cilium doesn't replace RDMA/InfiniBand for GPU collective traffic — that's a different network entirely.
vs / alongside
| Tool | Role | Note |
|---|---|---|
| Cilium | eBPF CNI for K8s networking | Management plane, CNCF Graduated |
| Calico | CNI with iptables or eBPF | Mature alternative, different approach |
| Multus | Multi-NIC pods | Adds secondary interfaces alongside Cilium |
| RDMA | GPU-to-GPU data plane | Different layer entirely |
References
- Cilium documentation — official docs.
- cilium/cilium — source (CNCF Graduated).
- eBPF datapath — how it works under the hood.
Extra reads
- Isovalent blog — deep dives from Cilium creators.
- ClusterMesh — multi-cluster networking.
Verified against Cilium docs (docs.cilium.io), May 2026.
Where Cilium fits: the mental model
Cilium is a foundational infrastructure service that moves, stores, connects, or distributes AI assets. The useful question is not simply “can it run the demo?” It is whether the component gives your team a clear ownership boundary, predictable failure behavior, and enough evidence to operate changes safely. Treat it as one replaceable layer in a larger system rather than letting it quietly become the architecture.
Start by drawing the request and data path. Mark where untrusted input enters, where identity is checked, where durable state changes, and where retries can repeat work. That diagram tells you which guarantees belong to Cilium and which still belong to your application, platform, cloud provider, or database. The distinction matters during incidents: a healthy process is not proof that the end-to-end task is correct.
Core concepts you should understand first
The vocabulary below is more important than any single SDK method. It lets application engineers, platform engineers, security reviewers, and incident responders describe the same system without confusing a framework feature with an end-to-end guarantee.
| Concept | Meaning in this layer | Design question |
|---|---|---|
| Data plane | The hot path that carries bytes, packets, objects, or artifacts. | Write down how Cilium represents or enforces this before production. |
| Control plane | APIs and controllers that configure, place, authorize, and observe the data plane. | Write down how Cilium represents or enforces this before production. |
| Consistency | What a reader may observe during concurrent writes, replication, or failure. | Write down how Cilium represents or enforces this before production. |
| Locality | Keeping compute near data or devices to reduce latency, egress, and cross-zone traffic. | Write down how Cilium represents or enforces this before production. |
| Identity | A workload or human principal used to authenticate and authorize every operation. | Write down how Cilium represents or enforces this before production. |
| Recovery objective | The measured RPO and RTO for metadata and data, not merely the presence of replicas. | Write down how Cilium represents or enforces this before production. |
From quick start to a production deployment
The earlier quick start proves that the package or service runs. Production readiness is a different exercise. Build the smallest vertical slice that crosses every real boundary—identity, network, persistence, upstream provider, telemetry, and rollback—before broadening the feature set.
- Pin the compatibility envelope. Record the Cilium release, language/runtime version, client SDK version, model or backend version, and—where applicable—Kubernetes API or driver requirements. Use a lock file, immutable image digest, or chart version; floating “latest” tags prevent repeatable rollback.
- Define contracts before configuration. Write the accepted input, successful output, error classes, timeout, idempotency behavior, and ownership of durable state. Validate at the boundary so corrupt work fails early instead of surfacing deep in a workflow.
- Create separate development, staging, and production identities. Do not copy a broad personal API key into every environment. Prefer workload identity or short-lived credentials, scope access by tenant and operation, and verify denial cases as part of deployment.
- Add bounded failure behavior. Every remote call needs a deadline. Retry only transient, idempotent operations with exponential backoff and jitter. Set concurrency and queue limits so an upstream slowdown becomes controlled backpressure rather than resource exhaustion.
- Instrument the complete path. Emit a correlation ID, component and release version, duration, outcome, retry count, and resource or cost dimensions. Keep sensitive prompt, document, and credential values out of ordinary logs.
- Ship through a reversible rollout. Run compatibility and regression tests, deploy to a canary or isolated workload, compare service-level indicators, then increase exposure. Preserve the previous artifact and configuration until rollback has been exercised.
Production configuration checklist
- Pin artifacts by version and, where possible, digest.
- Set connect, request, and total workflow deadlines.
- Bound retries, concurrency, queue length, and payload size.
- Separate read-only operations from mutations.
- Use idempotency keys for replayable mutations.
- Persist canonical state outside disposable workers.
- Encrypt traffic and durable data with managed keys.
- Redact secrets, tokens, prompts, and personal data.
- Apply per-tenant quotas and authorization filters.
- Expose readiness separately from process liveness.
- Back up metadata and test restore, not only backup.
- Document owner, escalation path, RPO, and RTO.
Failure modes and the response you should design
| Failure mode | What you observe | Engineering response |
|---|---|---|
| Metadata loss | Data exists but indexes, configuration, or ownership are gone. | Back up metadata separately and test restore to an isolated environment. |
| Cross-zone cost | A correct design creates unexpected egress and latency. | Make placement and traffic locality visible in cost and SLO dashboards. |
| Credential leak | Static secrets are copied into images or manifests. | Use workload identity, rotation, scoped roles, and secret scanning. |
| Capacity cliff | A quota, inode, object count, route, or device limit is reached. | Alert on forecasted exhaustion and document hard limits. |
| Split configuration | Nodes run incompatible policy or protocol versions. | Use staged rollouts and explicit version-skew rules. |
| Untested restore | Backups succeed but cannot recreate a working service. | Run scheduled restore drills and measure RPO/RTO. |
Turn these rows into runbook entries with an alert, first diagnostic query, safe mitigation, and escalation owner. Test at least one failure in staging every release cycle. If the system cannot be forced into a failure safely, it is usually not yet observable or isolated enough.
Security, privacy, and tenant isolation
Place Cilium in a threat model, not just an architecture diagram. Identify human users, workload identities, administrators, upstream services, model providers, artifact registries, and data stores. For each edge, document authentication, authorization, encryption, audit evidence, and the consequence of credential compromise.
Apply least privilege at the operation and resource level. A component that only retrieves documents should not be able to delete the index; an evaluation worker should not inherit production mutation credentials; a model-serving pod should not need cluster-admin. In multi-tenant systems, enforce the tenant boundary before retrieval or execution and include tenant identity in quotas and audit events. Never rely on a prompt instruction, namespace string supplied by the client, or UI filtering as authorization.
Decide what data is permitted in telemetry. Prompts, retrieved chunks, tool arguments, model responses, notebooks, and traces can contain secrets or regulated data. Redact close to collection, keep high-sensitivity payload capture opt-in, encrypt exports, restrict support access, and give each class an explicit retention period. Verify deletion across caches, replicas, indexes, backups, and derived evaluation datasets.
Observability and service-level objectives
A useful dashboard follows the user-visible unit of work and then decomposes it by component, release, tenant tier, backend, and failure class. Start with these signals for Cilium:
- availability and error rate — graph both rate and distribution, then compare with the previous release and traffic mix.
- p50/p95/p99 latency — graph both rate and distribution, then compare with the previous release and traffic mix.
- throughput and saturation — graph both rate and distribution, then compare with the previous release and traffic mix.
- replication or synchronization lag — graph both rate and distribution, then compare with the previous release and traffic mix.
- capacity and growth rate — graph both rate and distribution, then compare with the previous release and traffic mix.
- recovery time in drills — graph both rate and distribution, then compare with the previous release and traffic mix.
Choose an SLO at the boundary your users experience, such as “99% of accepted tasks complete correctly within five minutes over 28 days.” Availability alone is insufficient for AI systems because a fast but incorrect or ungrounded result is still a failure. Pair latency and completion objectives with a reviewed quality or policy indicator. Page on rapid error-budget burn; use tickets for slow capacity trends.
Testing and release strategy
Use four layers. Unit tests cover deterministic adapters, schemas, policy, and error mapping without a live external service. Contract tests exercise the pinned integration boundary—API, CLI, SDK, protocol, or ephemeral service—and verify its exact surface. Scenario tests exercise representative end-to-end cases, including permissions and state. Load and resilience tests establish saturation, queue behavior, retry amplification, and recovery after dependency loss.
Keep a small blocking suite for every commit and a broader scheduled suite for expensive or probabilistic checks. Store results with the application version, Cilium version, configuration hash, model/backend version, dataset version, and random seed. A score without that provenance cannot explain a regression. Before upgrading, read the migration notes, run both versions against the same replay set, and explicitly test rollback across any schema or state transition.
How to decide whether Cilium is the right tool
| Question | Evidence to collect | Red flag |
|---|---|---|
| Does it remove a real constraint? | A measured bottleneck, missing guarantee, or repeated custom component. | Adoption is based only on a demo or feature count. |
| Can the team operate it? | Named owner, upgrade path, alerts, runbooks, backup, restore, and on-call skills. | Only the original prototype author understands failure behavior. |
| Is the interface portable? | Your domain contracts wrap vendor-specific APIs; data and state have an export path. | Business objects are inseparable from framework internals. |
| Does it meet the envelope? | Benchmarks using your payloads, concurrency, topology, quality bar, and cost model. | Published benchmark hardware or workload does not resemble production. |
| Is failure affordable? | Tested degraded mode, bounded blast radius, rollback, RPO, and RTO. | A component outage blocks unrelated tenants or irreversible actions. |
Prefer the smallest component that satisfies the required guarantees. A provider SDK, relational table, background job, or standard Kubernetes controller is often better than another platform when the workload is small and predictable. Choose Cilium when its specific abstraction removes sustained engineering work and the team is willing to own its lifecycle.
A focused 90-minute validation lab
- Minutes 0–15: run the documented quick start in a disposable environment with pinned dependencies. Save the exact commands and a known-good input/output fixture.
- Minutes 15–35: replace the toy input with one representative case from your system. Add schema validation, a deadline, and a correlation ID.
- Minutes 35–55: force invalid credentials, a timeout, malformed input, and one dependency failure. Record the observed errors and whether retries are safe.
- Minutes 55–75: run a small concurrency test and capture latency, throughput, saturation, and unit cost. Do not extrapolate beyond the tested range.
- Minutes 75–90: write the adoption decision: required guarantees met, open risks, owner, next experiment, and the simplest credible alternative.
Frequently asked questions
Should we standardize on Cilium for every team?
Standardize the contracts, telemetry, security controls, and release evidence first. Standardizing one implementation is useful only when workloads share requirements and a platform team owns upgrades and support.
Can we use the hosted version and skip operations work?
Hosted service removes part of the control-plane burden, not architecture ownership. You still own identity, tenant isolation, data classification, quotas, dependency failure, observability, export, and an exit plan.
What should be pinned for reproducibility?
Pin the tool/server, client SDK, runtime, configuration, model or backend, container image digest, and test dataset. Record these values with every benchmark and evaluation result.
When is a proof of concept ready for production?
After representative success and failure tests pass, sensitive data paths are approved, limits and SLOs are defined, telemetry and runbooks exist, restore or rollback is rehearsed, and an accountable owner accepts the remaining risk.
Official sources and freshness
This guide was reviewed for architecture and operational guidance on 10 July 2026. Projects evolve quickly: verify installation syntax, supported versions, feature maturity, and upgrade notes against the exact release you deploy.