Pricing, SLAs, and cloud benefits
CapEx vs OpEx
CapEx is the big upfront hardware spend, the on-prem way. OpEx is paying for consumption monthly, the cloud way. Cloud shifts CapEx to OpEx, and consumption-based pricing means metered billing per second, minute, hour, or GB.
The cost savers: Reserved Instances (a 1 or 3 year commit, up to 72% off), Spot VMs (unused capacity at a deep discount, evictable on short notice, batch and test only), free tier services.
SLA math
The downtime-per-year ladder:
- 99%: 87.7 hours
- 99.9%: 8.76 hours
- 99.95%: 4.38 hours
- 99.99%: about 53 minutes
- 99.999%: about 5 minutes
Composite SLA: multiply the services, never average. App Service 99.95% x SQL 99.99% = 99.94%.
More dependencies, lower overall availability. Classic trick question.
Credits exist when Microsoft misses an SLA, but you must request them. And 99.99% usually signals availability zone support: deploy across zones to get there.
Serverless preview
Serverless means no servers to manage, billed per execution, nothing while idle. Azure Functions is code, event-driven; Logic Apps is a visual workflow designer with connectors. Functions = code, Logic Apps = designer; the exam loves that split.
Containers vs VMs: VMs carry a whole OS, containers share the host kernel and start in seconds. More in the compute notes.
Benefit vocabulary
High availability, scalability, elasticity, reliability, predictability (confidence in performance and costs), agility, security, governance, global reach, manageability.
Match the keyword to the scenario: holiday-traffic cost optimization is elasticity; control and compliance is governance.