In-Vehicle Container Orchestration & Mixed-Criticality Workloads
Bringing modern cloud-native deployment patterns inside the vehicle without compromising functional safety. Edge orchestrators schedule untrusted infotainment microservices alongside real-time telemetry pipelines, strictly enforcing CPU quotas and memory ceilings.
Transforming Vehicles into Ephemeral Cloud Nodes with Deterministic Pod Scheduling
Modern automakers deploy thousands of software engineers pushing daily code updates. Monolithic flash image updates are too risky and bandwidth-intensive for weekly vehicle upgrades. By deploying lightweight OCI container runtimes (such as crun) managed by edge Kubernetes distributions (such as K3s or Eclipse Kanto), OEMs push individual microservice patches over cellular networks without disrupting driving operations.
Core Engineering Areas
- Edge Container Orchestration — Standard Kubernetes (k8s) is designed for data centers with infinite RAM, drawing over 1GB of memory just for control plane components. Automotive architectures deploy stripped-down distributions like K3s or lightweight orchestrators like Eclipse Kanto. By replacing Docker/containerd with the lightweight C-based `crun` runtime, container initialization overhead drops to under 15 milliseconds.
- Mixed-Criticality Pod Scheduling — Preventing non-critical entertainment apps from degrading steering and braking safety microservices using hierarchical cgroups v2 bandwidth controllers.
- A/B Dual-Bank OTA Deployments — Architecting failure-proof flash storage layouts: passive bank verification, cryptographic signature validation, and sub-second rollback during boot failures.
Technical Articles