Design for high availability
High availability for compute, relational data, and semi-structured and unstructured data.
- 14 min
- 3 concepts
- Try it yourself
- 4-question check
Learn
High availability keeps a service running through component failures without manual recovery. The usual pattern is redundancy at each layer: across fault domains, zones, and for the most critical systems, regions. Composite SLAs multiply, so every dependency matters.
- 1
Recommend a high availability solution for compute
Spread VMs across availability zones (99.99%) behind a zone-redundant Standard Load Balancer or Application Gateway v2; use Virtual Machine Scale Sets with flexible orchestration across zones. For PaaS, use zone-redundant App Service plans, Container Apps environments and AKS node pools across zones. For regional resilience, deploy active-active or active-passive in two regions behind Azure Front Door or Traffic Manager.
Remember: Zones for datacentre failure; Front Door across regions.
- 2
Recommend a high availability solution for relational data
Turn on zone redundancy for Azure SQL Database or Managed Instance. Use active geo-replication (SQL Database, up to four readable secondaries) or failover groups (SQL Database and MI, a listener endpoint that doesn't change on failover). For SQL on VMs, use Always On availability groups across zones.
Remember: Failover groups keep the connection string the same.
- 3
Recommend a high availability solution for semi-structured and unstructured data
Storage: ZRS for zone failures, GZRS / RA-GZRS for region failures with readable secondary. Cosmos DB: enable multiple regions, availability zones and optionally multi-region writes (99.999% for reads and writes); service-managed failover moves the write region automatically.
Remember: Storage zone + region resilience: GZRS.
Composite SLA example
| Component | SLA |
|---|---|
| App Service (zone-redundant) | 99.95% or higher |
| Azure SQL Database (zone-redundant, Business Critical) | 99.995% |
| Both together | Multiply: lower than either alone |
| Two regions active-active | Higher than one region |
Try it
See which failures each compute design survives.
This interactive part needs JavaScript.
In the portal
- Zone redundancy for SQLAzure portal: SQL databases > your database > Settings > Compute + storage
- Failover groupsAzure portal: SQL servers > your server > Data management > Failover groups
- Front DoorAzure portal: Front Door and CDN profiles > Create
Key terms
Click a card to see the definition.
Check your understanding
4 questions. Score 80% or more and these skills are marked as confident in your AZ-305 study plan.
Which keeps the same connection string after an Azure SQL regional failover?
Failover groups provide a stable endpoint.
Two services with 99.95% and 99.99% SLAs depend on each other. The composite SLA is…
Multiply: 0.9995 × 0.9999 ≈ 99.94%.
Which gives global load balancing for HTTP across regions?
Front Door routes users to healthy regions.
Which storage redundancy survives both a zone and a region failure?
Zones in the primary plus a secondary region.
Sources
- Microsoft Learn: Design for high availability
- Microsoft Learn: Failover groups overview
- Microsoft Learn: Reliability in Azure Cosmos DB
- Microsoft Learn: Composite SLAs
Facts in this lesson are from Microsoft Learn. This site isn't affiliated with Microsoft.