Cost Optimization Scenarios
Part of the Cloud Computing Roadmap
In short
"Our cloud bill doubled last month. What do you do?" There's no trick answer. They want a calm, step-by-step investigation instead of "switch everything to spot instances".
'Our cloud bill doubled, how would you investigate?' — testing Phase 4 and Phase 12's cost optimization and monitoring knowledge applied to a diagnostic scenario.
Think of it like this
Who left the geyser on?
The electricity bill at home suddenly doubles. You don't sell the fridge in panic. You check the meter, compare with last month, and walk room to room. Ah, the new geyser has been on all day. Fix that first, then think about LED bulbs.
Watch it happen
Investigating a bill that doubled
1/6Is it real? Check for a one-time charge, a yearly commitment or support fee landing this month.
The big ideas
What you need to know
Tags make costs visible
Tag resources by team, project and environment. Without tags, "who owns this?" takes days to answer.
Data transfer hides
Traffic between zones, regions and out to the internet is billed per GB. It often explains a mystery spike.
Quick wins vs long wins
Deleting waste is fast. Right-sizing, commitments and re-architecture take longer but save more over time.
Catch it early
Budgets and anomaly alerts turn a month-end shock into a message on day two.
See it in code
Walk through the code
1# Unattached EBS volumes: still billed, used by nobody2aws ec2 describe-volumes --filters Name=status,Values=available \3 --query "Volumes[].[VolumeId,Size,CreateTime]" --output table4 5# Elastic IPs not attached to anything (still billed)6aws ec2 describe-addresses \7 --query "Addresses[?AssociationId==null].PublicIp" --output text8 9# NAT gateways: hourly charge plus a per-GB processing charge10aws ec2 describe-nat-gateways \11 --query "NatGateways[].[NatGatewayId,State,VpcId]" --output table1/3Volumes with status available are attached to no instance. Snapshot if unsure, then delete.
Three classic waste checks. In an interview, naming these shows you've actually looked at a real bill.
Save yourself the pain
Common mistakes
Oops
Jumping to "buy reserved instances" before knowing what caused the spike.
Do this instead
Find the cause first. Commitments discount steady usage; they don't fix a forgotten resource.
Oops
Deleting resources you think are unused without checking owners.
Do this instead
Check tags and ask owners, or snapshot first. Saving money by deleting prod data is a bad story.
Your mission
Practise the investigation
Quick check
The bill jumped on the 12th. Line items show NAT gateway data processing went up 10x. A likely cause?
Go deeper
- 1Practice a structured approach: identify the biggest cost drivers first, then investigate each.
- 2Prepare to discuss right-sizing, reserved pricing, and storage tiering as concrete levers.
- 3Practice presenting a cost optimization recommendation with an estimated savings figure.
More architecture & design at this stop
Cloud Architecture Design Questions
High Availability & DR Questions
Stuck on this topic? Ask an Insider
Get 1:1 guidance from people who've walked this exact path — on the InsideEdge app.
