Lumen Learn: A 15-Minute SLA Versus a Flat Budget
Lumen Learn is an edtech company with 150 employees and a monthly AWS bill of about $38,000. Roughly 60% is a steady-state production web tier and database; the remaining 40% is development and training environments that run hard for two weeks each quarter and sit mostly idle otherwise. Last month a production database issue took the customer portal down for five hours; the company's Basic support case received an automated response and a human reply three days later. The CTO now wants a 15-minute response commitment for business-critical outages and a named technical contact. The CFO wants total AWS spend, including support, to stay flat or fall, and has asked whether reserved capacity or a support plan upgrade is the better use of the next dollar. Lumen Learn is a single AWS account today and has just acquired a smaller studio that already runs its own workloads in AWS.
1. Which support plan meets the 15-minute requirement, what does it cost at this spend level, and what would you recommend given the CFO's constraint?
Read the worked answer
Only Enterprise Support provides a 15-minute response target for a business-critical system down, plus a designated Technical Account Manager. At $38,000 a month it costs the greater of $15,000 or a tiered percentage of usage, so about $15,000 a month - roughly $180,000 a year, or 39% of the AWS bill.
Business Support at the same spend is roughly $3,000 a month. It provides 24/7 phone, chat and email access to Cloud Support Engineers, a one-hour response target for a production system down, the full set of Trusted Advisor checks, and Infrastructure Event Management for a fee. It does not provide 15 minutes, and it does not provide a named TAM.
The CTO's requirement and the CFO's constraint are in direct conflict, and the useful answer names that conflict. $12,000 a month buys a response-time commitment, not availability. If five hours of downtime last quarter cost less than $144,000 a year in revenue and goodwill, buy Business, spend a fraction of the difference on Multi-AZ, health checks and rehearsed failover so fewer things break, and revisit Enterprise when a contractual SLA makes it unavoidable. Prices change, so verify current published rates.
2. How would you restructure the 60/40 spend mix, and what does consolidated billing add?
Read the worked answer
Separate the two halves. The 60% steady-state production workload is the right target for a commitment: a one-year, no-upfront Compute Savings Plan sized to the floor of that usage, so the commitment is not stranded if the tier is later right-sized. Savings Plans are more flexible than Convertible Reserved Instances here because they apply across instance family, size and Region, and also to Lambda and Fargate.
Do not commit against the 40% that runs hard for two weeks a quarter - that is a shutdown and Spot problem, not a commitment problem. Schedule non-production environments off outside working hours with Instance Scheduler, and run training jobs on Spot.
Consolidate the acquired studio into AWS Organizations. Consolidated billing gives one bill, aggregates usage for volume tiers such as S3 storage pricing, and pools the Savings Plan commitment across accounts; sharing can be disabled per account if chargeback isolation matters more. Tag everything for cost allocation, and set AWS Budgets and Cost Anomaly Detection before the next surprise rather than after.
3. The acquired studio wants to keep its 30 accounts and its autonomy. What governance structure do you recommend?
Read the worked answer
Use AWS Organizations, and Control Tower if you want guardrails provisioned rather than hand-built. Model the studio as its own organizational unit with separate accounts for production, development and logging, so autonomy is real but blast radius is bounded.
Service control policies at the OU level enforce non-negotiables: deny disabling CloudTrail, deny leaving the organization, restrict permitted Regions, require encryption on storage services. Centralize logs and security findings in a dedicated security OU, and use IAM Identity Center with permission sets so the studio's engineers keep their own roles without creating new long-lived credentials. Billing stays consolidated, which is the CFO's lever, while chargeback runs through tags and Cost Explorer rather than separate invoices.
The judgement call is how much policy to impose on day one. Too much and the studio's delivery stalls and shadow IT appears; too little and the acquisition becomes the weakest link in the security posture. Start with logging, encryption and Region restrictions, then add the rest as the two teams' practices converge.