Modern enterprise workloads demand high availability, resilience, and automated recovery across geographically separated locations. VMware Cloud Foundation (VCF) addresses these requirements through Multi Availability Zone (AZ) deployments using vSphere and vSAN stretched clusters. This architecture enables continuous service availability even during an entire Availability Zone outage, while maintaining centralized management and operational simplicity.
What Is a Stretched Cluster in VCF?
A vSphere stretched cluster is a single vSphere cluster with vSAN storage that spans multiple Availability Zones (AZs). Each AZ is treated as an independent Failure Domain, typically representing a separate data center, building, or fault-isolated zone with independent power, cooling, and network paths.
In a stretched cluster architecture:
-
Compute hosts are distributed across two AZs.
-
vSAN synchronously replicates data between AZs.
-
A vSAN witness node provides quorum to prevent split-brain scenarios.
This design allows workloads to remain available and consistent even if one AZ becomes unavailable.
Key Architectural Components
1. Availability Zones as Failure Domains
Each Availability Zone acts as a vSAN Fault Domain, ensuring that data replicas are placed across AZs. This guarantees that the loss of an entire AZ does not impact data availability or integrity.
2. vSAN Stretched Storage
vSAN uses synchronous replication between AZs to maintain identical data copies. This ensures:
-
Zero data loss (RPO = 0)
-
Transparent storage access for virtual machines
-
Consistent performance under normal operating conditions
3. vSAN Witness Node
A vSAN witness node is mandatory for stretched clusters. It:
-
Maintains quorum during failures
-
Resides in a third fault location (separate from both AZs)
-
Stores only metadata, not workload data
The witness ensures deterministic behavior during AZ failures and enables automated recovery.
High Availability and Automated Recovery
One of the major benefits of stretched clusters in VCF is the retention of native vSphere features:
-
vSphere HA automatically restarts affected virtual machines in the surviving AZ in the event of an AZ failure.
-
DRS (Distributed Resource Scheduler) continues to balance workloads dynamically based on available compute resources.
-
No manual intervention is required for workload recovery, significantly reducing operational risk and recovery time.
This results in a fully automated and highly resilient platform capable of surviving complete AZ outages.
Deployment Order and Best Practices in VCF
Management Domain First
In VCF, the Management Domain cluster must be stretched first before stretching any VI Workload Domain (WLD) vSAN clusters. This is critical because:
-
The Management Domain hosts core infrastructure services such as SDDC Manager, vCenter, NSX, and lifecycle management components.
-
Ensuring its availability is foundational for the entire VCF stack.
Stretch VI Workload Domains as Needed
After the Management Domain is successfully stretched:
-
Additional vSAN-based VI Workload Domain clusters can be stretched selectively.
-
Not all clusters need to be stretched—only those hosting workloads with strict availability and resiliency requirements.
-
This approach allows organizations to balance cost, complexity, and availability.
Use Cases and Benefits
Multi-AZ stretched clusters in VCF are ideal for:
-
Mission-critical applications requiring continuous availability
-
Environments with zero or near-zero downtime requirements
-
Enterprises seeking simplified disaster avoidance rather than traditional disaster recovery
Key benefits include:
-
Protection against entire AZ failures
-
Zero data loss and rapid recovery
-
Fully automated operations using native vSphere capabilities
-
Consistent management through VMware Cloud Foundation
Conclusion
VCF Multi Availability Zone deployments using stretched clusters provide a powerful, resilient architecture for modern private cloud environments. By leveraging vSphere HA, DRS, and vSAN stretched storage—with the correct deployment order starting from the Management Domain—organizations can achieve enterprise-grade availability with minimal operational overhead.
This architecture transforms Availability Zones into active-active infrastructure components, ensuring business continuity even in the face of major infrastructure failures.
