Reliability and predictability
The next pair of cloud benefits is reliability and predictability.
Reliability
Reliability is a system's ability to recover from failures and keep functioning. Because the cloud is decentralised, with data centres in regions all over the world, a failure in one place does not have to take your application down.
You can design for reliability by:
- running multiple copies of an application across availability zones or regions,
- keeping backups and geo-replicated data,
- planning disaster recovery so you can fail over to another region.
Some Azure services do much of this for you automatically.
Predictability
Predictability comes in two flavours:
Performance predictability: you can be confident your application will have the resources it needs. Autoscaling, load balancing and high availability all help: if demand rises, more resources arrive; if a server fails, traffic moves to healthy ones.
Cost predictability: you can forecast and control what you will spend. Azure helps with the pricing calculator (estimate before you deploy) and Microsoft Cost Management (analysis, budgets, alerts and forecasts once you're running), which you meet later in the course. Committing to reservations or a savings plan makes the bill more predictable too.
The Well-Architected Framework
Microsoft's Azure Well-Architected Framework collects guidance for building good workloads around five pillars:
- Reliability
- Security
- Cost optimisation
- Operational excellence
- Performance efficiency
AZ-900 only needs you to know it exists and roughly what it covers.