Domain 4 is 12% of the scored content, the smallest of the four and the most factual. Most of it is three lists: the purchasing options, the cost tools, and the Support plans. Learn them by what distinguishes each entry from the one next to it.
Compute purchasing options
- On-Demand. Pay per second or per hour with no commitment. Unpredictable workloads, short-term work, development, anything new.
- Reserved Instances. A one- or three-year commitment to a specific instance configuration, up to 72% below On-Demand. Steady-state workloads with a known instance type.
- Savings Plans. A one- or three-year commitment to an hourly spend figure in US dollars. Steady usage where the instance family or Region may change.
- Spot Instances. Spare EC2 capacity at up to 90% below On-Demand, reclaimed on two minutes’ notice. Fault-tolerant, interruptible work: batch, rendering, CI, stateless web tiers.
- Dedicated Instances. Instances on hardware isolated to your account. A USD$2 hourly fee applies in every Region where you run at least one. Regulatory isolation requirements.
- Dedicated Hosts. A whole physical server, with visible sockets and cores. Per-socket or per-core licences, bring-your-own-licence, compliance that needs a known host.
- Capacity Reservations. Capacity held in one named Availability Zone. An immediate-use reservation carries no term commitment; a future-dated one commits for a duration you set. Neither carries a discount, so pair one with a Savings Plan or a Regional Reserved Instance.
Reserved Instance detail:
- Standard RIs give the largest discount and can be modified: Availability Zone, scope, and instance size within the family, that last only on Linux/Unix with default tenancy. They cannot be exchanged. Convertible RIs discount less, modify the same way, and can also be exchanged for another Convertible with a different family, operating system or tenancy.
- Payment options: All Upfront (largest discount), Partial Upfront, No Upfront (smallest).
- Scope: a Regional RI discounts usage in any Availability Zone in the Region and gives instance size flexibility within the family, but it reserves no capacity. A Zonal RI reserves capacity in one Availability Zone and gives neither Availability Zone nor size flexibility.
- In AWS Organizations, discount sharing is on by default, so an unused RI or Savings Plan discount in one account applies to matching usage in another. The management account can deactivate sharing per account under Billing preferences, which holds a commitment’s benefit in the account that purchased it.
- The Reserved Instance Marketplace lets you sell a Standard RI you no longer need. Registering as a seller requires the root user.
Savings Plans come in four kinds. Compute Savings Plans are the most flexible, covering EC2, Fargate and Lambda usage regardless of instance family, size, operating system, tenancy or Region, at up to 66% off On-Demand. EC2 Instance Savings Plans commit to one instance family in one Region for up to 72% off. Database Savings Plans cover Aurora, RDS, DynamoDB, ElastiCache and several other database services, up to 35% off. SageMaker AI Savings Plans cover SageMaker AI usage, up to 64% off. All four take the same three payment options. None of them cover Spot usage.
Spot behaviour: EC2 issues a two-minute interruption notice, then terminates, stops or hibernates the instance. A rebalance recommendation can arrive earlier, when the risk of interruption rises. A workload that cannot be interrupted is the wrong fit; a queue of restartable jobs is the right one.
Storage pricing shapes
- Amazon S3. Storage per GB per month by class, requests, a per-GB retrieval charge on the colder classes, and data transfer out.
- S3 minimum storage durations. 30 days for Standard-IA and One Zone-IA, 90 days for Glacier Instant Retrieval and Glacier Flexible Retrieval, 180 days for Glacier Deep Archive. Delete early and the full period is still charged. S3 Standard and Intelligent-Tiering have no minimum; Intelligent-Tiering charges a per-object monitoring fee and has no retrieval fee.
- Amazon EBS. Provisioned capacity per GB per month whether or not it is used, plus provisioned IOPS and throughput on some volume types. Snapshots are charged separately.
- Amazon EFS. What is actually stored, by storage class, with no capacity to provision.
- Instance store. Nothing separately; it is included in the instance price.
EBS is charged on what is provisioned, S3 and EFS on what is stored. So an over-provisioned EBS volume is a standing cost, and an empty bucket is nearly free.
Data transfer charges
- Into AWS from the internet. Free.
- Out of AWS to the internet. Charged per GB, after 100GB a month free, aggregated across services and Regions.
- Between Regions. Charged per GB.
- Between Availability Zones in a Region. Charged in both directions.
- Within one Availability Zone over private IPv4 addresses. Free.
- Out to the internet through CloudFront. Charged at CloudFront rates. Transfer from an AWS origin into CloudFront is free.
Inbound free and outbound charged is the shape to remember, along with cross-AZ traffic being charged even though it never leaves the Region.
Billing and cost management tools
- AWS Pricing Calculator. Estimates the cost of an architecture before it is built.
- The per-service pricing pages. Every service has one, and it is where the published rate lives. The Calculator estimates an architecture; a pricing page states a rate.
- The AWS Price List API. The same published prices, queryable, for anyone who wants them in a system rather than on a page.
- The AWS Free Tier. Three shapes worth separating: always free, twelve months free from sign-up, and short-term trials.
- AWS Cost Explorer. Visualises past and current spend, with forecasts and rightsizing recommendations.
- AWS Budgets. Sets a cost, usage, RI or Savings Plans target and alerts when it is approached or exceeded. It can trigger an action; it does not stop resources running.
- AWS Cost and Usage Report. The most detailed billing data available, delivered to S3 for analysis.
- AWS Billing Conductor. Customised billing views and rates, usually for chargeback to internal teams or end customers.
- AWS Cost Anomaly Detection. Machine learning over your spend, alerting on unusual movement.
- AWS Compute Optimizer. Recommends instance sizes and types from observed utilisation.
- AWS Trusted Advisor. Best-practice checks, including idle and underutilised resources.
Before or after separates the first two. The Pricing Calculator estimates what something will cost, Cost Explorer analyses what it did cost, and Budgets alerts on a threshold going forward.
AWS Organizations and cost allocation
- AWS Organizations. Central management of many accounts, grouped into organisational units under a root.
- Management account. The account that created the organisation. It pays the bill, and service control policies never apply to it.
- Consolidated billing. One bill for all accounts, with usage aggregated so volume discount tiers are reached sooner, and RI and Savings Plan discounts shared across accounts.
- Service control policies. Guardrails capping what any principal in a member account can do, that account’s root user included. They grant nothing themselves; permissions still come from IAM policies.
- AWS Control Tower. A governed landing zone built on Organizations, with account vending and pre-built controls.
Cost allocation tags attribute spend to a team, project, environment or cost centre. AWS-generated tags carry the reserved aws: prefix, aws:createdBy for example, and you cannot edit them. Tags you define appear in the cost allocation report under a user: prefix, and you have to activate the tag key in the Billing and Cost Management console first. The key can take 24 hours to appear on that page, and another 24 hours to activate.
Tagging late is only partly recoverable. A resource is never tagged retrospectively, so a tag applied today says nothing about last month’s usage. Where the tag was already on the resource and only the activation was missing, you can request a backfill covering up to the previous 12 months, starting on the first day of a month, once every 24 hours.
AWS Support plans
First-response targets run down a five-step severity ladder, and the ladder is the same whatever the paid plan: general guidance 24 hours, system impaired 12 hours, production system impaired 4 hours, production system down 1 hour, and business-critical system down in minutes. Basic Support carries no technical cases at all, so no ladder applies to it.
- Basic. Free with every account. Account and billing questions, service quota increases, documentation, whitepapers, re:Post, the AWS Health Dashboard, and a baseline set of Trusted Advisor checks. No technical support cases.
- Developer. Business-hours email access to Cloud Support Associates, general guidance, and a 12-business-hour target on an impaired system. General guidance and system impaired are the only severities it carries.
- Business. Under 1 hour for a production system down, 24/7 phone, email and chat with Cloud Support Engineers, the full Trusted Advisor check set, the AWS Health API, third-party software support, and unlimited contacts.
- Enterprise On-Ramp. Under 30 minutes for a business-critical system down, a pool of Technical Account Managers, a Concierge team, and consultative architectural review.
- Enterprise. Under 15 minutes for a business-critical system down, a designated Technical Account Manager, Concierge, well-architected and operations reviews, training, and AWS Countdown event management.
The lineup is being retired. Developer, Business and Enterprise On-Ramp are closed to new customers and AWS discontinues all three on 1 January 2027, except in AWS GovCloud (US), where they remain. What AWS sells now is Basic, Business Support+ (from USD$29 a month, 24/7 engineers, more than 500 Trusted Advisor checks, under 30 minutes on a business-critical case), Enterprise (from USD$5,000 a month, reduced from USD$15,000, designated TAM, under 15 minutes, AWS Security Incident Response at no extra charge) and AWS Unified Operations (from USD$50,000 a month, adding Incident Management Engineers and a five-minute business-critical response). The CLF-C02 guide still names the older four, so hold both in your head.
Learn the response times, because they discriminate between tiers more reliably than anything else: 1 hour is Business, 30 minutes is Enterprise On-Ramp, 15 minutes is Enterprise. Two other lines separate the tiers. The full set of Trusted Advisor checks starts at Business, and a designated Technical Account Manager is Enterprise only, where Enterprise On-Ramp gave a pool.
AWS Trusted Advisor
Checks an account against best practice in six categories: cost optimisation, performance, security, fault tolerance, service limits, and operational excellence. Basic and Developer Support reach every check in the service limits category plus six named checks across security and fault tolerance, with no automatic updates, so the security ones need a manual refresh. Business and above reach all of them, more than 500 checks, through the console and the Trusted Advisor API. The exact count moves as AWS adds checks, so learn where the full set starts rather than the number.
Health, partners and technical resources
- AWS Health Dashboard, service health view. Public reporting on AWS services by Region, with 12 months of service history, readable without signing in.
- AWS Health Dashboard, account health view. Events affecting your resources, and your organisation’s accounts where Organizations is in use. Free to every customer.
- AWS Health API. Programmatic access to those events, from Business Support upward.
- AWS Partner Network. Consulting partners (systems integrators, resellers) and technology partners (independent software vendors).
- AWS Marketplace. Third-party software purchased through your AWS account, with cost management, governance and entitlement features. It appears on the AWS bill.
- AWS Professional Services. AWS’s own consulting organisation.
- AWS Solutions Architects. Technical guidance on architecture, as part of the account relationship.
- AWS re:Post. Community questions and answers, successor to the AWS Forums, and now the home of the AWS Knowledge Center articles that answer common support questions.
- AWS Prescriptive Guidance. Patterns, guides and migration strategies from AWS teams.
- AWS whitepapers and documentation. The official written material, including the Well-Architected Framework.
- AWS Trust & Safety. The team to contact to report abuse of AWS resources.
AWS IQ, which matched customers with AWS-certified freelancers, shut down on 28 May 2026, so it is not an answer to anything now; AWS Marketplace Professional Services covers that ground. Benefits of being an AWS Partner include training and certification, partner events, funding programmes, and volume discounts.
Traps
- Pricing Calculator is before, Cost Explorer is after. An estimate for an architecture not yet built is the Calculator.
- Budgets alert; they do not cap spend. A budget can trigger an action; by itself it does not halt a running resource.
- Savings Plans commit to a spend figure per hour; Reserved Instances commit to an instance configuration. A Compute Savings Plan keeps its discount through a change of instance family, and a Standard RI does not.
- A Regional RI reserves no capacity. Only a Zonal RI holds capacity, and only in its own Availability Zone. For held capacity without a term, use a Capacity Reservation.
- Spot is not a discount you can rely on. Two minutes’ notice, so an interruptible workload is a requirement rather than a preference.
- Dedicated Instances are isolated hardware; Dedicated Hosts show you the sockets and cores. A per-core licence needs a Host.
- Consolidated billing shares RI and Savings Plan benefit across accounts in the organisation, and aggregates usage so volume tiers arrive sooner.
- Inbound data transfer is free; outbound is charged after 100GB a month. Cross-Availability-Zone traffic is charged even though it stays in the Region.
- User-defined cost allocation tags have to be activated in the Billing console, and a resource is never tagged retrospectively.
- The full Trusted Advisor check set starts at Business Support. Basic and Developer get the service limits category plus a handful of security and fault tolerance checks, rather than none.
- A designated Technical Account Manager is Enterprise only. Enterprise On-Ramp gave a pool.
- Service health is public; account health is yours. A question about an event affecting your specific instances is the account view.
- Changing the Support plan is not on AWS’s root-only list, despite how often it is listed as one. IAM permissions govern it, though a service control policy cannot block signing up for Enterprise Support as the root user.