Unveiling Enhanced CPU Pods: Docker Runtime and Network Volume Support
We're excited to announce two significant upgrades to our CPU pods that will streamline your development workflow and expand storage options. Our CPU pods now feature Docker runtime (replacing Kata Containers) and support for network volumes—previously exclusive to our GPU pods.
Consistent Developer Experience
Using the same Docker runtime across local development and our cloud platform eliminates the "works on my machine" problem. Your containerized applications will behave consistently whether developed locally or deployed to our platform.
Improved Performance
Docker's native runtime delivers lower container startup times and reduced overhead compared to Kata Containers, resulting in faster deployments and more efficient resource utilization.
Enhanced Compatibility
The switch to Docker increases compatibility with standard container tooling and workflows, allowing seamless integration with popular CI/CD pipelines and development tools.
Important Note: While our previous Kata-based pods supported Docker-in-Docker (nested containerization), this capability is no longer available with our Docker runtime CPU pods. We recommend adjusting your workflows to pre-build your Docker images and store them in a container registry before deployment. This may also be a good time to explore our GitHub integration for serverless, if your CPU workloads are a good candidate for it. While existing setups will continue to work for the moment, the plan is to move all customers to Docker containers for CPU over time.
Network Volume Support: Flexible Storage Solutions
CPU pods now support network volumes, previously available only for GPU workloads. This brings several advantages:
Persistent Data Access
Network volumes provide persistent storage that exists independently of pod lifecycles, ensuring your data remains accessible even after pod termination or restart.
Seamless Data Sharing
Multiple pods can now mount the same network volume simultaneously, enabling efficient data sharing across workloads without complex data transfer mechanisms.
Cost-efficient pricing
Network volumes can be dynamically resized to accommodate growing storage needs without pod reconfiguration, providing flexibility for evolving workloads. You can save up to half on your storage bill by moving it to a cost-efficient network volume:
| Storage Type | Under 1TB | Over 1TB |
|---|---|---|
| Container or Disk Volume (running pod) | $0.10/GB/Month | $0.10/GB/Month |
| Disk Volume (stopped pod) | $0.20/GB/Month | $0.20/GB/Month |
| Network Volume | $0.07/GB/Month | $0.05/GB/Month |
Containers are wiped when pods are stopped, so there is no applicable charge.
Getting Started
These enhancements are available immediately for all CPU pod users. To leverage these new capabilities, configure network volumes through our dashboard.
We're confident these improvements will significantly enhance your development experience on our platform. As always, we welcome your feedback as you explore these new capabilities.