Skip to main content
Skip table of contents

Shared Costs

Shared costs are expenses that support multiple teams, applications, or customers and cannot be attributed to a single owner based on tags alone. Examples include shared Kubernetes clusters, networking, security tools, logging platforms, and enterprise support plans. This guide describes common approaches for handling shared and indirect costs in Cloudaware FinOps so that showback/chargeback remains fair and transparent.

Why Shared Costs Need Special Treatment

If shared costs are ignored or lumped into a generic bucket:

  • Some teams appear cheaper than they really are because platform or support costs are hidden.

  • Central platform teams cannot demonstrate the full cost of running shared services.

  • Business owners may perceive internal bills as arbitrary or unfair.

A clear approach to shared costs makes it easier to explain bills and drive the right behavior.

Identify Shared Cost Categories

Start by identifying major shared cost categories, such as:

  • Central networking and security services.

  • Shared Kubernetes or PaaS clusters.

  • Logging, monitoring, and observability platforms.

  • Central data platforms or shared databases.

  • Cloud provider enterprise support and account‑level fees.

Document which cost items belong to each category (for example, specific accounts, subscriptions, projects, or service names).

Allocation Methods for Shared Costs

Common methods include:

  • Do not allocate (central bucket)

    • Keep some costs in a central “Platform” or “Shared Services” bucket.

    • Often used for early‑stage programs or costs that are intentionally not passed through.

  • Fixed percentage splits

    • Split a shared category across a small number of BUs or products based on agreed percentages.

    • Simple and predictable, but may need periodic adjustment.

  • Usage‑based allocation

    • Allocate costs in proportion to a measurable driver (for example, CPU hours, storage used, number of pods, API calls).

    • Aligns costs more directly with consumption patterns and encourages efficient use.

  • Per‑unit or per‑user allocation

    • Divide costs by a business unit such as number of users, customers, or transactions.

    • Useful for SaaS or multi‑tenant offerings.

Using Effective Cost and Commitments

For providers like AWS, Cloudaware exposes effective cost metrics that amortize commitments (RIs, Savings Plans, reservations) over the resources that benefit from them. When allocating shared commitment costs:

  • Prefer effective cost metrics so that consumers see the true economic cost of their usage.

  • Avoid double‑counting by deciding whether to allocate raw on‑demand cost, amortized commitment cost, or both.

Using Split Cost Allocation Data

Some cloud providers, such as AWS, support split cost allocation as an optional feature that must be enabled explicitly. For Amazon EKS, split cost allocation introduces separate Cost and Usage columns (e.g., splitLineItem/SplitCost, splitLineItem/NetUnusedCost) for container-level workloads. These split cost values roll up to the parent EC2 instance’s Amortized Cost.

Cloudaware supports split cost allocation metrics and can build analytics dashboards based on this data, enabling customers to analyze Kubernetes and container costs with greater granularity.

Implementing Shared‑Cost Decisions

In Cloudaware FinOps:

  • Use your business mapping model to route shared accounts, subscriptions, or projects to the appropriate shared‑cost categories.

  • Apply allocation rules or calculations in dashboards and reports or downstream BI Tools that implement the chosen method per category.

  • Keep a small number of well‑named categories so stakeholders can understand them quickly.

Revisit shared‑cost assumptions periodically as your architecture and organizational structure evolve.

For how these decisions show up in internal bills and statements, see Invoicing & Chargebacks.

JavaScript errors detected

Please note, these errors can depend on your browser setup.

If this problem persists, please contact our support.