Azure SLAs and Composite SLAs

A Service Level Agreement (SLA) is a formal commitment from Microsoft defining the minimum uptime and connectivity guarantee for each Azure service. SLAs are expressed as a percentage — for example, 99.9% means the service will be available at least 99.9% of the time in a given month. If Microsoft fails to meet the SLA, customers are eligible for service credits as compensation.

Each Azure service has its own SLA document published at azure.microsoft.com/en-us/support/legal/sla/. When reading an SLA, pay attention to: the uptime percentage, the conditions required to achieve it (e.g., deploying across two or more instances), and the credit table that specifies how much credit you receive for each level of underperformance.

An SLA percentage translates to an allowed downtime per month:

• 99.9% = up to 43.8 minutes downtime per month
• 99.95% = up to 21.9 minutes per month
• 99.99% = up to 4.38 minutes per month
• 99.999% = up to 26 seconds per month

Understanding these numbers helps you assess whether a service's SLA meets your business continuity requirements.

When your application depends on multiple Azure services, each with its own SLA, the overall availability is the product of all individual SLAs. This is called the composite SLA. If any one service is unavailable, the application is unavailable, so the composite SLA is always lower than the lowest individual SLA in the chain.

Consider an application using Azure App Service (99.95%), Azure SQL Database (99.99%), and Azure Cache for Redis (99.9%). The composite SLA is: 0.9995 × 0.9999 × 0.999 = 0.9984 = 99.84%. This means the application could be unavailable for up to 2.4 hours per month, even though each individual service has a strong SLA. Adding more services in a chain increases this risk.

You can improve your composite SLA by adding redundancy — running parallel instances of a service so that if one fails, the other continues. For services in parallel, the formula changes: the combined unavailability is the product of each service's unavailability. For example, two App Service instances each at 99.95% in parallel gives: 1 - (0.0005 × 0.0005) = 99.9997%.

Some Azure services have no SLA when used in a free or shared tier. For example, Azure App Service on the Free (F1) or Shared (D1) tier has no uptime guarantee. This is acceptable for development and testing, but production workloads must use a paid tier to qualify for an SLA. Always check the SLA conditions before deploying production resources.

Azure Virtual Machines have different SLAs depending on the configuration. A single-instance VM with Premium SSD gets a 99.9% SLA. The same VM deployed in an availability set gets 99.95%. Two VMs deployed across availability zones reach 99.99%. This demonstrates that architecture choices directly affect the SLA you can claim and commit to your own customers.

When Microsoft fails to meet an SLA, customers can claim service credits — percentage discounts applied to future bills. Credit amounts increase with the severity of the breach: typically 10% credit for uptime between 99.0% and 99.9%, 25% for below 99.0%, and 100% in some cases. Service credits are the only remedy — Microsoft does not pay cash compensation for SLA breaches.

Microsoft's SLA is a guarantee about service availability, not about your application's specific Recovery Time Objective (RTO) or Recovery Point Objective (RPO). Even if a service meets its SLA, you may still need additional architecture patterns — like multi-region failover or regular backups — to meet your own business continuity targets. The SLA is a floor, not a ceiling.

Use Azure Monitor and Azure Service Health to track whether services are meeting their SLAs. Azure Service Health provides notifications about service outages, planned maintenance, and health advisories that may affect your resources. Set up health alerts to be notified immediately when an Azure service in your region has a problem.

Test your understanding of Microsoft Azure Fundamentals (AZ-900) concepts from this lesson.

In this lesson you learned: an SLA is a formal uptime guarantee from Microsoft expressed as a percentage; the composite SLA for services in series is calculated by multiplying all individual SLAs; and redundancy in parallel can dramatically improve the composite SLA. Next up we explore availability sets and availability zones to understand how to achieve high SLAs for VM workloads.