A commitment to a level of uptime, such as 99.99% for VMs spread over availability zones, 99.95% for an availability set and 99.9% for one VM.
Also called service level agreement.
Read more: Microsoft Learn
In the Ultra Transcenders books
AZ-305AI-300AZ-900AI-200DP-900AZ-400AZ-802DP-800ALZ
Each book explains SLA in context, with comparison tables and the common traps.
Terms in this definition
- Virtual machines
Infrastructure-as-a-service compute giving complete control of the operating system, making it a fit for lift-and-shift moves and for software relying on OS-level pieces like COM.
- OVER
Gives a T-SQL window function its window: PARTITION BY, ORDER BY and, if wanted, a ROWS or RANGE frame. Rankings and running totals can then be worked out while every row is kept.
- Availability zones
Separate physical datacentre locations inside one region. Placing VMs in at least two of them earns a 99.99% SLA.
- Availability set
Grouping that distributes VMs across as many as 3 fault domains and 20 update domains inside a single datacentre, backed by a 99.95% SLA. It offers neither autoscale nor zone distribution.
Related terms
- Azure DDoS infrastructure protection
The no-cost DDoS defence that Azure applies automatically at its network edge to every public IP address, with nothing to set up. Unlike the paid tiers, it comes without a customer SLA, attack analytics or rapid response support.
- Azure Spot Virtual Machines
Heavily discounted VMs using Azure's spare capacity. Azure can reclaim them with 30 seconds' warning when it needs the capacity back or the price rises above your limit, and with no SLA they fit work that can tolerate interruption.
- Bounded staleness
Cosmos DB consistency setting where reads may trail writes, but never by more than a configured count of versions or time interval; it is the strongest level that carries a write-latency SLA for multi-region accounts.
- Case management
Tracks SecOps work in the Defender portal as cases, either incident or generic, complete with SLA policies, attachments, comments, tasks and a history of changes. Sentinel in the Azure portal lacks it.
- Developer deployment type
Deployment type meant purely for trying out fine-tuned models; each deployment lasts 24 hours and comes with no SLA or data residency commitment, so it isn't suitable for production.
- GA
The release stage that follows preview, when a feature becomes fully supported and its SLA applies.
- Global Batch
An asynchronous deployment type for big files of requests, aiming to finish within 24 hours at roughly half the price, with its own quota. With no real-time SLA, it is unsuitable for interactive scenarios.
- Microsoft Entra Health
An area of Microsoft Entra for monitoring service health, giving near-real-time signals and anomaly alerts for scenarios like MFA and compliant-device sign-ins (preview), plus monthly SLA attainment. Viewing it requires Reports Reader or higher.