AZ-305 Designing Microsoft Azure Infrastructure Solutions Exam

Recommend a high availability solution for compute

High availability in compute ensures that applications keep running even when individual virtual machines fail. It relies on redundancy and fault isolation to avoid single points of failure. Azure offers several services to distribute compute resources across separate physical locations. These designs aim to meet strict uptime targets and minimize unplanned downtime.

One common approach is to use Availability Sets. These sets group VMs into multiple fault domains and update domains. Fault domains spread VMs across different racks and power sources, while update domains ensure that planned maintenance doesn’t take all VMs offline at once.

• Fault domains protect against hardware failures.
• Update domains handle rolling updates with minimal service interruption.

For even greater isolation, Azure provides Availability Zones, which are distinct datacenters within the same region. Each zone has its own power source, network, and cooling. By deploying VMs in two or three Availability Zones, you achieve resilience to datacenter-level failures. This setup is ideal for mission-critical workloads that require the highest uptime.

When workloads must adapt to changing demand, Virtual Machine Scale Sets (VMSS) offer both high availability and automatic scaling. VMSS can span Availability Zones or sets and integrates with Azure Load Balancer. It automatically scales out or in based on performance metrics. This combination of fault tolerance and elasticity ensures consistent performance.

Recommend a high availability solution for relational data

Designing high availability for relational databases focuses on keeping data accessible and consistent. Azure provides managed services that handle much of the complexity. Key strategies include automatic failover, geo-replication, and redundant storage.

For Azure SQL Database, Active Geo-Replication allows up to four readable secondary databases in different regions. These secondaries stay continuously synchronized with the primary. In case of an outage, you can manually or automatically fail over to a secondary to restore service quickly.

• Failover Groups extend this concept by grouping databases for coordinated failover.
• Automated health monitoring triggers a failover if the primary becomes unhealthy.

Azure Database for MySQL and PostgreSQL offer Zone Redundant Configuration. They replicate data synchronously across Availability Zones in the same region. This provides near-zero data loss and fast automatic failover. Built-in backups ensure you can restore to any point in time.

For on-premises or custom database engines, Azure Site Recovery can orchestrate replication and failover to Azure. It continuously copies transactions from your primary server to Azure VMs. If your main site fails, Site Recovery performs a planned or unplanned failover. This approach extends high availability beyond a single datacenter.

Recommend a high availability solution for semi-structured and unstructured data

High availability for non-relational data in Azure focuses on Storage Accounts, Cosmos DB, and Blob Storage. The goal is to ensure data is always available, even if a datacenter or region goes offline. Replication plays a central role in this design.

Azure Storage offers multiple replication options:

• Locally Redundant Storage (LRS) replicates data within one datacenter.
• Zone-Redundant Storage (ZRS) spreads data across three zones in a region.
• Geo-Redundant Storage (GRS) replicates asynchronously to a paired region.
• Read-Access GRS (RA-GRS) adds read access to the secondary region.
  Choosing the right option balances cost, performance, and resilience.

For globally distributed NoSQL data, Azure Cosmos DB provides turnkey high availability. It supports multi-master writes and multi-region replicas with single-digit millisecond latency. Consistency levels range from strong to eventual, so you can choose how up-to-date reads must be. Cosmos DB automatically handles failover based on your configured priority of regions.

When working with large files or unstructured datasets, Azure Data Lake Storage Gen2 builds on Blob Storage’s redundancy features. It integrates with analytics engines like Azure Databricks and HDInsight. By combining zone or geo-replication with hierarchical namespaces, you get both reliability and performance. This is ideal for big data workloads that need to stay online.

Conclusion

Designing for high availability in Azure compute involves spreading virtual machines across Availability Sets, Availability Zones, and using Virtual Machine Scale Sets. These strategies ensure that both planned maintenance and unexpected hardware failures have minimal impact on services. Automatic scaling and load balancing further enhance reliability and performance.

For relational databases, leveraging Active Geo-Replication, Failover Groups, and Zone Redundant Configuration in Azure SQL Database, MySQL, and PostgreSQL provides continuous data protection and fast failover. Azure Site Recovery extends these capabilities to custom database engines and on-premises setups, maintaining business continuity across sites.

Semi-structured and unstructured data achieves high availability through the choice of replication options in Azure Storage, global distribution in Cosmos DB, and integration in Data Lake Storage Gen2. By selecting the appropriate replication strategy and consistency model, you can balance cost, performance, and resilience to keep data accessible under any conditions.