Cedana
cedana.aiAutomation Layer for AI Factories
Ops & Infragpucontainerscheckpoint-restorehpcai-infrastructurecomputekubernetes
About
Cedana is a Save/Migrate/Resume (SMR) platform for containerized CPU and GPU workloads. It sits between the Linux kernel and your workloads to checkpoint and migrate container state across instances and cloud vendors. The system is designed to maximize compute utilization, improve reliability, and support policy-based orchestration for AI and HPC workloads.
Problem
Compute resources are wasted due to idle instances and workloads that can't seamlessly migrate across nodes or cloud vendors.
For
AI and HPC engineers managing containerized workloads
How it works
Cedana intercepts between the Linux kernel and containers to save the full state of a workload, enabling live migration and resumption across different instances and vendors via an API.
Business model
unknown
Status
launched
Company
Cedana