A Guide to Cloud Cost Visibility

Companies have migrated to the cloud for the ability to access on-demand computing resources and pay only for the resources used. And yet, the majority of cloud-based companies are overspending on cloud. In this guide to FinOps, we’re explaining how you can set up effective cloud cost management strategy across your organization, and reduce one of your biggest line items.

The Inform Phase involves setting up visibility into your organization’s cloud spend in order to inform future optimization decisions.

We’ll start by explaining the importance of forming a dedicated FinOps team. Then we’ll look at how to set up your cloud accounts for data extraction to create source-of-truth cost dashboards that measure relevant efficiency metrics. Finally, we’ll explain how the FinOps team can use these dashboards for budgeting and alerts, and for mapping out appropriate approvals processes.

‍

1. Form a FinOps team

Look to your left, now look to your right. If you don’t see someone running FinOps, guess what: you’re running FinOps.

If your enterprise is spending 6 figures on cloud computing, it might be worth sharing the load and establishing a team to take accountability for org-wide cloud budgeting. You can do this either by upskilling existing team members or by hiring additional talent dedicated entirely to managing this team.

FinOps is a cross-functional effort that requires collaboration among several key players. This multidisciplinary team should include representatives from all departments that have a stake in cloud cost management

‍

‍

This team should start with two goals: effectively setting up cloud spend monitoring systems and cloud spend approvals processes.  These two should then be continuously reviewed and adapted as the team begins to meet at a regular cadence.

‍

2. Set Up Cloud Spend Monitoring

Monitoring cloud spend involves setting up data workflows from your various cloud spend data sources  (from your CSP, a third party management tool, or just from an excel sheet someone has been manually updating over time)  into a single source of truth FinOps dashboards.  

‍

Start with your CSP’s public cloud cost data. Public cloud pricing data contains the details about a cloud provider’s rates for its services. Combine this with your public cloud billing data, which contains detailed, granular consumption and billing information including cloud account identifiers, data, time, price, cost, and technology types. Acquiring this dataset involves ingesting data from your various cloud providers using multiple mechanisms, such as retrieving CUR files from AWS S3 buckets for object storage or utilizing APIs for Microsoft Azure, among others.

Next, process and clean the data. Tags act as labels that tie cloud resources (EC2 instances, S3 buckets, AMIs etc.) to specific departments, projects, or teams. Tagging policies and third party tools can help with maintaining and normalizing tag data; especially for multi-cloud organizations. Not all tags on cloud resources appear in the cloud billing data, so gathering the full set of tags associated with cloud resources can add further context. Implement a clear and consistent naming convention across resources, and enforce these through tagging policies or automated tagging scripts. Each tag comprises a key and a value, both of which you generate, e.g.:

• ^{owner = engineering}
• ^{project = marketingWebsite}
• ^{environment = production}
• ^{sensitiveData = true}
• ^{application = fin-app-1}
• ^{resize = false}
• ^{businessUnit = 016}

Use the final cost data to create BI dashboards. All this granular data allows you to then create detailed dashboards and reports, which are critical for visualizing cloud cost data and extracting actionable insights. There are a few options:

• Use in-house cloud tools like AWS’s Cost Explorer and then use a BI tool to extract cost data into open source dashboards like CUDOS. This integrates directly into your cloud instance, and means dedicated support from cloud service providers. However, it can be difficult to customize how data is analyzed and displayed.
• Extract all the data from your cloud service provider entirely and run it through generic third-party data tools like Tableau or Power BI. This offers highly customizable data displays and almost full control over what to do with them. This is not dedicated to cloud data so requires a lot of setup and maintenance, and you are responsible for building all the pipelines and workflows.
• Use third party cloud specific data programs like Datadog, CloudZero, or New Relic to gain cost visibility. This will still require time and effort to set up and integrate, and will be an additional cost and contract to manage. But it will be easier than creating your own pipelines to build customizable dashboards.

Make sure you’re measuring efficiency. Effective cloud cost optimization ensures that cloud costs correlate with productive and profitable activities. Cloud is an investment intended to increase business value, and increasing cloud spend is sometimes just a reflection of the company doing well. The goal of the FinOps team is therefore not necessarily to reduce total spend but to ensure spend efficiency.

To do this requires setting up continuous monitoring of relevant spend efficiency metrics.

When setting up your dashboards, here are some efficiency metrics that you need to ensure you are tracking:

Review these dashboards at regular audits. The FinOps team can now gain ROI insights such as cost per customer, cost per feature, and cost per deployment. For finance managers, dashboards can reveal spending trends and top cost drivers. For engineering managers, they reveal historical usage trends and enable better forecasting for Reserved vs Spot instances. Make it a joint effort with engineering to set a schedule for  periodic audits to examine this cloud usage data and tagging standards line by line. Use these reviews to update your inventory of both current and past cloud services and their status and provide qualitative context to usage trends, or unused/underutilized resources. 

‍

3. Establish a FinOps Framework

Your FinOps team's second priority is to  establish clear, organization-wide policies for cloud usage. Using the FinOps dashboards, you can now  analyze past spending patterns to inform realistic budgets and benchmarks..

Without proper checks, it’s easy for individuals to spend money spinning up new cloud resources or subscribe to additional services without considering the broader impact on the budget. Implement a robust approvals system for new purchases with approval tiers based on spend levels and justification using metrics on estimated cost, duration and ROI of the resources.

Determining appropriate benchmarks for cloud usage means you can now configure alerts to notify your team when expenses approach or exceed these pre-set limits, allowing for timely corrective action. For instance, alerts can trigger automatic reviews of high-spending resources or initiate scale-down processes for underutilized assets. You can use tools like Harness CCM to detect statistical cost anomalies for your Kubernetes clusters and cloud accounts (both one-time cost spike, and gradual or consistent cost increases).

Example Policies:

• Requiring ownership and reporting labels so containers can be cost-allocated
• Requiring workloads to have CPU and memory requests set to improve efficiency and reliability
• Only permitting certain EC2 instance types for certain teams
• Providing feedback to developers on the cost impact of their resource changes

Example Policy enforcement tools:

• Kubernetes admission controllers
• Policy engines (Polaris and Open Policy Agent, for example)
• Continuous Integration and Continuous Deployment (CI/CD) pipelines to automate policy and guardrail enforcement when developers make pull requests.
• Vendr and Omnea procurement request and approvals automation

You can implement policies at each stage of the software development lifecycle.  An effective FinOps practitioner, however, will be mindful, however, of not bogging teams down with bureaucracy, and minimize or automate approvals for repeat or standard purchases.

                                                                     ***

‍

Cloud-based teams can provision IT infrastructure with the click of a button, free from IT controlling their budget. This cuts down overhead, but it also removes accountability - managing cloud costs is never going to be an individual team’s priority.

By forming a cross-functional centralized FinOps team, you now have a unit responsible for organization wide cost management and financial planning. This involves setting up  monitoring dashboards, and enforcing consistent tagging policies and regular inventory audits.  With the help of these dashboards, the team can then forecast and allocate budget more effectively, and set up reasonable alerts and approvals policies.

You will likely find that having a dedicated team armed with finops tools and an  accurate source-of- truth dashboard will lead to some  immediate one-time fixes. You can now:

• Identify major cost contributors and then cut sources of idle compute or excess spending
• Identify where workload leads to higher earnings and then devote more resources there

However, you will soon uncover that most sources of inefficient spending require more advanced, continuous optimizations. This is because the FinOps model is a game of real-time adjustments to your business’s changing needs. Read on for Phase 2, where we explain how to do this.

‍

‍