AZ-305 Designing Microsoft Azure Infrastructure Solutions Exam

Recommend a High Availability Solution for Compute

Architect Zone-Redundant Compute Deployments

Availability Zones and Availability Sets are two critical constructs in Azure for ensuring high availability in compute deployments. Availability Zones consist of physically separate datacenters within a region, each equipped with independent power, cooling, and networking. They provide fault isolation from entire datacenter failures. On the other hand, Availability Sets group virtual machines (VMs) across multiple fault domains and update domains within a single datacenter, thereby guarding against faults at the rack level.

Virtual Machine Scale Sets allow you to deploy and manage a set of identical VMs with integrated autoscaling and health monitoring features. When configuring a zone-redundant scale set in flexible orchestration mode, Azure ensures VM instances and their disks are evenly spread across several zones. Key configuration aspects include:

• Zone balance: Use best-effort for flexibility or strict for maintaining exact VM count parity across zones.
• Fault domain spreading: Opt for max spreading to maximize resilience or align with managed disk fault domains.
• Health probes: Utilize Azure Load Balancer to exclude unhealthy VMs from rotation.

In scenarios not utilizing scale sets, individual VMs and disks can still be distributed across three Availability Zones to achieve redundancy. Each VM should have its OS and data disks assigned to specific zones, managing traffic with an Azure Load Balancer. To ensure resilience, it is essential to:

• Provision at least two VMs per zone.
• Attach zone-redundant disks (ZRS) to prevent data loss.

Safe deployment and rolling upgrade practices are crucial for continuous service during platform and application updates. Prior to initiating node upgrades in a new zone, it is recommended to:

• Monitor the Azure Service Health Dashboard for any status updates and planned maintenance.
• Employ load balancer health probes to drain traffic from the ongoing upgrade zone.
• Execute rolling upgrades one zone at a time to maintain operational capacity.

Autoscaling policies with VM Scale Sets help sustain performance and availability during fault conditions. By defining rules such as "CPU > 80%" or custom metrics, instances can be automatically scaled up or down across zones. The integration of Azure Load Balancer and health probes ensures that new instances receive traffic only after successful health checks, guaranteeing uninterrupted compute services even during failures.

Conclusion

In summary, creating a high availability solution for compute within Azure involves strategically utilizing Availability Zones, Availability Sets, and Virtual Machine Scale Sets. Distributing workloads across multiple zones protects against datacenter outages while spreading VMs within a zone guards against localized failures. Essential practices include balancing zonal deployments, using redundant disks, conducting safe rolling upgrades, and implementing autoscaling based on health metrics. Adopting these techniques ensures resilient and continuous compute services, minimizing impact during various failure scenarios.