Dedicated Host (DDH) gives you a single-tenant environment with exclusive access to all physical resources on a cloud host — CPUs, memory, and network interface controller — backed by Alibaba Cloud virtualization technology. Use DDH when your workloads require physical isolation for security and compliance, license portability, or predictable performance.
Isolate your workloads for security and compliance
On a DDH, no other tenant shares your physical resources. Every dedicated host carries a unique machine code, which helps you meet strict security and compliance requirements across industries and enterprise environments.
Deploy and move instances with full control
Flexible placement
When creating an ECS instance, specify a dedicated host directly, or let the system assign one automatically. With the host association feature, an ECS instance stays on the same physical server even when stopped — preserving a consistent deployment layout across your environment.
Flexible migration
Move ECS instances between two dedicated hosts, or between a shared host and a dedicated host. This lets you rebalance workloads or respond to maintenance needs.
Reduce costs with overprovisioning and BYOL
CPU overprovisioned host type
For workloads with low CPU utilization — such as development, testing, and system O&M — use the CPU overprovisioned dedicated host type to customize the vCPU-to-memory ratio and pack more instances per host. Deploying ECS instances on a dedicated host can cost up to 48% less than traditional deployment methods. For available host types, see Dedicated host types.
Bring your own license (BYOL)
If your software licenses are assigned by CPU cores, CPU sockets, and number of virtual machines, continue using them on a dedicated host under the same license terms. BYOL lowers the total cost of migrating existing workloads to the cloud.
Economical mode
Enable economical mode on an ECS instance to release its CPU and memory resources when the instance is stopped.
Keep services running with automatic failover
Enable automatic instance migration so that if a dedicated host fails, its ECS instances are migrated to another host. Combine this with load balancing to further improve service availability.