Five nines across any clouds
Run Compute Anywhere™
20+ pre-configured regions on AWS, GCP, Azure, and OCI, run by Control Plane. No clusters or cloud accounts to set up. Turn a location on and workloads deploy there in minutes.
Control Plane can provision and run CNCF-certified Kubernetes inside your AWS, GCP, Azure, Hetzner, or Linode account. Or bring your own. Register an existing EKS, GKE, AKS, or any conformant cluster as a location with one kubectl command.
Any SSH-accessible Linux server can join a cluster as a node. Generate a join script from the console or CLI, run it on the machine, done. Servers in a data center, a store, a factory, or a colo become a region of the same cloud, with the same networking, identity, and observability as everything else.
- Bare metal
- Virtual machines
- AWS
- Microsoft Azure
- GCP
- OCI
- IBM Cloud
- Gcore
- Alibaba Cloud
- RunPod
- DigitalOcean
- Crusoe
- Akamai (Linode)
- Vast.ai
- Hetzner
- Hyperstack
- OVHcloud
- Leaseweb
- Vultr
- Open Telekom Cloud
- CoreWeave
- Infomaniak
- Lambda
- NAVER Cloud
- Nebius
- Rackspace Technology
- Scaleway
- HPE GreenLake
- Huawei Cloud
- Dell APEX
- Tencent Cloud
- Latitude.sh
- Civo
- phoenixNAP
- UpCloud
- Fujitsu
- Exoscale
- NTT
- IONOS Cloud
- Any other cloud
Shape and reshape your cloud
Locations are switched on and off per GVC. Enable one and workloads are provisioned there; disable it and they are removed. Each GVC can have a different set of locations.
GVC ProductionLocations 4
A global cloud
- Geo-routing. Every workload gets a TLS endpoint served from all of its locations. Traffic goes to the nearest healthy location by latency-based DNS geo-routing.
- Automatic failover. If a location goes down, or a whole provider does, traffic moves to the next healthy location. Per-GVC routing options set priority-based failover, latency offsets, and latency thresholds when “nearest” is not the rule you want.
- Any workload type. Standard, stateful, cron, and serverless containers; virtual machines; sandboxes for AI agents.
- Everything else, built in. Service mesh, firewalls, and Cloud Wormhole; Universal Cloud Identity and 80+ deployable services; zero-trust policy, secrets, and audit; logs, metrics, and tracing.
Scaling
- Per workload, per location. Set minimum and maximum replicas and a scaling strategy (concurrent requests, requests per second, CPU utilization, request latency at a chosen percentile, memory utilization, or KEDA metrics).
- Scale to zero. Serverless workloads scale to zero on request-based strategies; standard and stateful workloads scale to zero on KEDA metrics.
- Capacity AI. Right-sizes CPU and memory for each workload automatically, so you stop paying for headroom you never use.
- Cluster scaling. Managed Kubernetes clusters in your accounts autoscale their node pools to match demand.