AWS, Azure & GCP · Cloud across providers
What cloud pricing models should you know?
Short answer
On-demand pays by the second or hour with no commitment; savings plans and reserved capacity trade commitment for discounts; spot uses spare capacity at a steep discount but can be interrupted. Storage and egress add cost.
Cloud pricing is usage-based at its core, but providers layer discounts and risk transfer on top. Choosing a model is a forecast problem: you are trading flexibility for a lower unit price, and the discount is only real if utilisation matches the commitment.
The main models
- On-demand: maximum flexibility, highest unit price, good for spiky or uncertain workloads
- Reserved capacity: commit to a specific instance family or resource for one or three years
- Savings plans: commit to spend per hour, often more flexible across instance families and regions
- Spot or preemptible: use spare capacity at a large discount, with interruption notices and no guarantee
- Committed-use discounts and sustained-use discounts vary by provider and service
Hidden costs often dominate the bill. Data egress, cross-AZ traffic, NAT gateway processing, load balancer hours, request charges and storage retrieval fees are easy to miss. A workload that looks cheap on compute can be expensive once network and requests are counted.
How to choose
- Cover only the stable baseline with commitments, and leave spikes on on-demand
- Use spot for fault-tolerant batch, stateless workers and CI runners
- Model the break-even utilisation before signing a multi-year commitment
- Re-evaluate when architecture or instance families change
Follow-ups you should expect
Interviewers use these to check whether the first answer was understanding or memorisation.
- 1How do you decide how much baseline to cover with commitments?
- 2What workloads are safe to run on spot?
- 3Which network charges surprise teams most often?
- 4How do savings plans differ from reserved instances?
Say it in the interview
The main models are on-demand for flexibility, reserved capacity or savings plans for a discount in exchange for commitment, and spot for cheap but interruptible capacity. I cover only the stable baseline with commitments and keep spikes on on-demand. I also model egress, cross-AZ traffic and request charges, because those often decide whether a design is actually cheap.
Use this as a model, not a script. Saying it in your own words is what makes it survive follow-up pressure.
Related questions
Reading is step one
Now say it out loud.
Understanding an answer and delivering it under pressure are different skills. Rehearse this topic in a live AWS, Azure & GCP mock interview and get scored feedback on depth, clarity and confidence.