Global digital service delivery
Deliver websites, APIs, downloads, and dynamic traffic with an edge layer that protects the application and a compute foundation that stays close to the workload.
Compute
Cloud compute with a model for every workload.
Cloud servers powered by our self-developed virtualization platform, with compute, network, and storage configured on demand. Choose balanced performance, dedicated resources, or discounted interruptible capacity for business systems, backend services, background jobs, and staging environments.
Instance models
Performance Instance
A balanced default for general workloads.
Dedicated Instance
Stronger isolation for steady workloads.
Spot Instance
Lower cost when the workload can be reclaimed.
Choose the resource model first.
Start with how your workload behaves, then choose the capacity model that matches it.
| Instance | Best for | Resource model | Pricing approach |
|---|---|---|---|
| Performance | Everyday business systems and backend services | Flexible shared capacity | Standard instance pricing |
| Dedicated | Workloads that need predictable performance | Assigned vCPU and memory | Dedicated resource pricing |
| Spot | Batch jobs, CI, and fault-tolerant services | Interruptible spare capacity | Market-driven discounts |
One virtualization layer. More room to choose.
Cloud Servers keep the infrastructure surface simple while giving each workload a suitable operating model.
Self-developed virtualization
A virtualization layer built and operated by ExtraLink.
Workload-aware scheduling
Compute resources are configured around the workload you need to run.
Compute, network, and storage
Configure the core resources together as the deployment changes.
Scale with the plan
Scale manually, with plan-based auto-scaling where available.
A practical fit for the systems you already run.
Persistent services, operational jobs, and regional environments can share the same compute foundation.
01
Business systems and internal services
02
Backend services and worker processes
03
Staging and preview environments
04
Regional workloads
Configuration at a glance.
Set the resources around the service, environment, or job you need to run.
- Instance types
- Performance, dedicated, and spot instances
- Virtualization
- Self-developed virtualization platform
- Storage
- SSD and NVMe options
- Network
- Configurable bandwidth with public and private networking
- Scaling
- Manual scale-up and plan-based auto-scaling
Questions, answered.
What instance types are available?
Cloud Servers provide three options: Performance Instances for balanced everyday workloads, Dedicated Instances with isolated compute resources for predictable performance, and Spot Instances for interruptible workloads at market-driven discounts.
What is the difference between Performance and Dedicated Instances?
Performance Instances are designed for general-purpose workloads with flexible shared capacity. Dedicated Instances reserve the assigned vCPU and memory resources for your workload, providing stronger isolation and more predictable performance.
How do Spot Instances work?
Spot Instance pricing follows available capacity and demand, so the price is lower and can change over time. Instances may be reclaimed when capacity is needed; use them for fault-tolerant workloads such as batch jobs, CI, data processing, and stateless services.
Can I run background jobs?
Yes. Cloud Servers are suited to persistent backend services, worker processes, batch jobs, and operational workloads.
Powered by Cloud Servers.
Global game operations
Keep game services, APIs, downloads, live events, and player traffic responsive across regions, with acceleration and security at the edge and compute behind the service.
Multi-region enterprise network
Connect offices, services, and regional environments through carrier-grade private routes with multi-route acceleration and compute placed around each business system.
AI application platform
Give AI applications one model access layer, route requests across providers, and run the surrounding services on cloud compute with global connectivity when required.
Production AI agent operations
Move from model calls to repeatable agent workflows: unify model access, orchestrate tools, and run task execution on dependable service infrastructure.