Cloud benefits - scalability, elasticity, availability and SLAs
The exam likes near-identical terms. This lesson separates them so the right word jumps out of a scenario.
Scaling and elasticity
Scalability is the ability to add or remove resources to meet demand.
- Vertical scaling (scale up/down): change the size of a machine, for example more CPU or RAM.
- Horizontal scaling (scale out/in): change the number of instances, for example from 2 to 6 web servers.
Elasticity is scaling that happens automatically as demand changes. Azure autoscale rules and Virtual Machine Scale Sets are typical examples. Scalable means you can; elastic means it happens by itself.
Other benefits
| Benefit | Plain meaning |
|---|---|
| Agility | Deploy and change resources quickly to respond to new requirements |
| Predictability | Costs (pricing calculator) and performance can be forecast |
| Reliability | The system recovers from failures and keeps running |
| Security | Built-in tools, compliance certifications, and threat protection |
| Governance | Policies, standards and compliance applied consistently |
| Manageability | Manage resources through the portal, templates, automation and monitoring |
| Consumption-based pricing | Pay only for what you use, no up-front capital |
Note the pairing of high availability (the service stays up with minimal downtime) and fault tolerance (continuing to operate despite a failure). In Azure, both are achieved by spreading resources over availability zones and regions.
Service level agreements
A service level agreement (SLA) is Microsoft’s commitment to the uptime of a service, expressed as a percentage per month. Higher numbers mean less allowed downtime:
| SLA | Approximate downtime per month |
|---|---|
| 99% | about 7 hours |
| 99.9% | about 43 minutes |
| 99.95% | about 22 minutes |
| 99.99% | about 4 minutes |
Uptime is calculated as the share of the month the service was available. If Microsoft misses the SLA, customers can claim a service credit, not a refund of the whole bill.
You can improve an app’s availability by deploying to several instances or zones. Services in a chain multiply their SLAs, which lowers the total. Redundant parallel paths raise it.
Lifecycle of a service
Azure services can be in preview (public or private) before general availability (GA). Preview services usually carry no SLA and are not recommended for production.
Exam traps
- Adding RAM to one VM is vertical scaling, not elasticity.
- Elasticity is automatic. Manual scaling is only scalability.
- High availability and disaster recovery are not the same: HA limits downtime, DR restores after a major outage.
- A dependent chain of services has a lower composite SLA than any single member.
- Preview features generally have no SLA.
Practice questions
Choose an answer, then review the explanation to check your understanding.
1. A VM is upgraded from 4 to 16 GB of memory to handle more load. What is this called?
Show answer
Answer: B.Adding CPU or memory to an existing machine is vertical scaling (scale up). Adding more instances is horizontal scaling.
2. A shop's web app automatically adds instances during a sale and removes them afterward. Which benefit is shown?
Show answer
Answer: A.Elasticity means resources grow and shrink automatically with demand, so you pay only for what is needed.
3. Workloads must keep running if one datacenter in a region fails. Which concept is this?
Show answer
Answer: C.Fault tolerance is the ability to keep working when a component or datacenter fails, for example by using multiple availability zones.
4. Two services in a chain each have a 99.9% SLA. What is the composite SLA when both are required?
Show answer
Answer: D.Multiplying the SLAs (0.999 x 0.999 = 99.8%) gives a lower composite SLA, because either failure breaks the solution.