We’re pleased to share the July 2026 Container Services newsletter, bringing together recent updates across OCI’s container and serverless portfolio. Like our April edition, this issue includes blogs covering OCI Kubernetes Engine (OKE), Container Instances, OCI Functions, and self-managed Kubernetes integrations. To make the updates easier to scan, we’ve organized them around practical platform themes: larger cluster scale, additional networking and security controls, better lifecycle management, more useful observability, and smoother developer workflows.
OKE now supports larger and more specialized workloads
These OKE scale updates focus on larger clusters, faster node-to-node communication, and better lifecycle support for dedicated bare metal capacity.
OKE now supports clusters with up to 20,000 worker nodes, giving customers a larger single-cluster option when their workloads benefit from one scheduler, one large capacity pool, and fewer cluster boundaries. This is especially relevant for large-scale AI/ML and GPU workloads, batch and high-churn jobs, and distributed data-processing environments where spreading capacity across many smaller clusters can add operational complexity.
Read more: https://blogs.oracle.com/cloud-infrastructure/oke-now-supports-clusters-with-up-to-20000-nodes
OKE added managed node support for OCI Compute Clusters with RDMA. This helps customers run RDMA-capable workloads on managed node pools rather than relying only on self-managed worker nodes for fast node-to-node communication. For distributed training, fine-tuning, multi-node inferencing, and HPC-style workloads, lower-latency communication between supported bare metal instances can help customers make better use of accelerator capacity.
Read more: https://blogs.oracle.com/cloud-infrastructure/oci-kubernetes-engine-rdma-compute-clusters
For customers using dedicated bare metal capacity, OKE now supports Compute Host Groups and Quick Recycle for managed nodes. This can reduce how long dedicated bare metal capacity remains unavailable after worker node termination, depending on configuration. The update is especially relevant when customers need managed node-pool lifecycle benefits while also using dedicated capacity for infrastructure-intensive workloads.
Read more: https://blogs.oracle.com/cloud-infrastructure/oci-kubernetes-engine-host-groups-quick-recycle
OKE adds more networking and security controls
These updates give teams more control over how workloads connect, receive traffic, and enforce communication policy.
Generic VNIC Attachment (GVA) for OKE is now generally available. GVA gives finer control over pod networking by enabling multiple secondary VNICs per node pool, each with its own subnet, network security groups, and IP allocation settings. GVA also increases the supported number of pods per node. With the new GVA capability, teams can design workload isolation, traffic segmentation, and IP address usage around their application and security needs.
Read more: https://blogs.oracle.com/cloud-infrastructure/gva-ga
OKE Managed Load Balancers now support cookie-based session persistence for Services of type `LoadBalancer`, and OCI Native Ingress Controller supports the same capability for ingress-based traffic. Customers can choose between application cookie persistence and load balancer cookie persistence depending on how the application handles session state. This is useful for applications that need predictable session behavior across repeated client requests.
Read more: https://blogs.oracle.com/cloud-infrastructure/cookie-persistence-oke-lb
OKE now supports Zero Trust Packet Routing (ZPR) for supported cluster-related OCI resources. ZPR lets teams assign security attributes to supported resources and write policies that define approved communication paths. Existing network, IAM, Kubernetes RBAC, and NetworkPolicy controls remain important; ZPR adds another policy check for supported OCI resources with security attributes.
Read more: https://blogs.oracle.com/cloud-infrastructure/zpr-support-for-oke
Day-2 operations continue to improve
Lifecycle and operations continue to improve.
Practical guide to export logs and metrics from OCI Container Instances shows two ways to centralize observability for Container Instances workloads: a sidecar collector pattern that forwards shared log and metric files with no application code changes, and a direct OCI SDK approach for teams that want native OCI Logging and OCI Monitoring integration with a leaner runtime. The blog gives customers a practical path to move Container Instances workloads toward production-ready troubleshooting and operations without giving up the service’s simplicity.
Read more: https://blogs.oracle.com/cloud-infrastructure/oci-container-instances-logging-solution
New OKE observability add-ons improve visibility into data plane health. The ObservabilityAgent add-on collects infrastructure metrics from sources such as kubelet, cAdvisor, CoreDNS, and kube-proxy. The NodeProblemDetector add-on surfaces node-level issues such as resource saturation and network problems. Together, these add-ons can help customers and OKE operators detect and troubleshoot worker-node issues sooner.
Read more: https://blogs.oracle.com/cloud-infrastructure/oci-kubernetes-engine-observability-add-ons
Virtual Node Cycling gives customers a simpler way to keep existing Virtual Nodes aligned with the latest Virtual Node Pool configuration. Customers can update the Virtual Node Pool, then trigger cycling to replace existing Virtual Nodes with new ones that inherit the desired configuration. Rollout controls such as `maximumSurge` and `maximumUnavailable` help manage the pace of replacement.
Read more: https://blogs.oracle.com/cloud-infrastructure/virtual-node-cycling
OKE now supports Kubernetes v1.36. This release gives OKE customers access to current upstream Kubernetes features, including fine-grained kubelet API authorization, user namespaces, mutating admission policies, and dynamic resource allocation.
Read more: https://blogs.oracle.com/cloud-infrastructure/kubernetes-v1-36-is-now-available-on-oke
OKE also now supports Oracle Linux 9 worker nodes. OL9 brings newer UEK kernels, OpenSSL 3, updated SELinux userspace packages, cgroup v2 support, and newer container tooling. Customers can use OL9 images for new OKE node pools and plan upgrades for existing environments that need the newer OS, security, and runtime foundation.
OCI expands cloud-native management, development, and operations
Recent updates expand how teams manage OCI resources from Kubernetes, develop and observe Functions, and use persistent file storage with Container Instances.
OCI Functions Skills add AI-guided help for the practical work of serverless development on OCI, including setup, deployment, validation, and troubleshooting. The new deploy and troubleshoot skills help users navigate OCI auth, Fn context, OCIR access, logs, metrics, and common failure paths with clearer next steps and safer actions.
Read more: https://blogs.oracle.com/cloud-infrastructure/oci-functions-ai-serverless-deployment-skills
Crossplane Provider for OCI now supports Crossplane v2. The update adds support for namespaced managed resources, reconciliation controls, observe-only behavior, and updated examples and migration guidance. For platform teams, these capabilities can help align OCI resource management with Kubernetes-native control-plane patterns.
Read more: https://blogs.oracle.com/cloud-infrastructure/crossplane-provider-for-oci-crossplane-v2
OCI Service Operator for Kubernetes (OSOK) and kro can be used together to define and deploy an application stack through a Kubernetes-native workflow. The example shows how Kubernetes resources and OCI infrastructure services can be orchestrated together, reducing the need to manage application and infrastructure layers separately.
Read more: https://blogs.oracle.com/cloud-infrastructure/manage-k8s-apps-on-oci-with-osok-and-kro
OCI Functions local development has also improved. New local debug support lets developers attach a debugger while invoking a function locally, and `fn watch` automatically redeploys after code changes. Together, these updates shorten the edit-build-invoke-repeat loop for local Functions development.
Read more: https://blogs.oracle.com/cloud-infrastructure/debug-support-hot-reload-for-functions-local-dev
OCI Functions now supports structured logging. Customers can configure applications so JSON-formatted log lines remain structured in OCI Logging, making it easier to view fields, add columns, filter by values, and search application logs.
Read more: https://blogs.oracle.com/cloud-infrastructure/structured-logging-in-oci-functions
OCI Container Instances now support persistent file storage through File Storage Service, making it easier to run stateful and file-based workloads without giving up the simplicity of serverless containers. Customers can mount durable shared storage directly into container instances to preserve data across restarts, replacements, and new runs, including enterprise patterns such as subdirectory mounts, read-only access, encryption in transit, and same-region cross-tenancy access.
Read more: https://blogs.oracle.com/cloud-infrastructure/announcing-fss-support-for-container-instances
Closing
Collectively, these July updates reflect continued investment across the OCI cloud-native platform, including Kubernetes infrastructure, networking, security, operations, developer tooling, and Kubernetes-native integrations. OKE is expanding support for larger and more specialized infrastructure, adding more networking and security controls, and improving lifecycle and operational visibility. OCI Functions continues to improve developer and operational workflows. Kubernetes-native integrations such as Crossplane, OSOK, and kro help teams manage applications and infrastructure through familiar Kubernetes patterns.
Explore the linked posts for implementation details, release guidance, and examples your teams can apply.