GUIDE · PRIVATE CLOUD

FinOps for hybrid infrastructure: how to keep costs under control

This guide shows you how to bring order to the costs of an infrastructure that mixes in-house servers, public cloud services and software licences. After reading it you will know how to gather the right costs, assign them to the people who use them, spot the most common waste and set up a periodic review that leads to concrete decisions.

Updated october 2026
7 min read
Altovar Technical Editorial Team

1. Why hybrid costs get out of hand

In a hybrid infrastructure, costs come from different sources: the purchase of servers, maintenance contracts, software licences, energy and technical-room space, public cloud charges and staff time. Each item is managed by a different person or provider, often with a different document. The result is that nobody sees the total cost of a service, and decisions are made on the most visible item rather than the largest one.

FinOps, from the combination of finance and operations, is a method that connects financial management with technical management so that costs become visible, attributable and governable. It is neither software to install nor an automatic spending cut. It is a repeated process that links the numbers to decisions, and it works only when finance and technical teams work from the same data.

2. Gather all cost items

The first step is to build a cost inventory, even if it is imperfect. Include everything needed to run a service, not just public cloud invoices. The cost of hardware should be treated according to the accounting rules your company applies, to be checked with your accountant, whereas fees and support are usually easier to read. Do not forget people's time: it is often the largest item and the least tracked.

For each item note the source, the billing frequency, the service it supports and the owner. An inventory with declared gaps is more useful than a complete one that has been invented. Clearly mark estimated items so that nobody treats them as firm data when the budget is discussed. Update the inventory every time a service is activated or retired.

  • Purchase and depreciation of servers and network equipment, under the accounting rules adopted.
  • Maintenance, support and extended warranty contracts.
  • Software licences tied to processors, machines or users.
  • Energy, cooling and technical-room space, even if estimated.
  • Public cloud fees and usage, with their ancillary charges.
  • Time of internal staff and of providers dedicated to the infrastructure.

3. Allocate costs to those who use them

A cost that is not attributed to anyone is never discussed. Allocation links each item to a service, and the service to a department or project. Labels, known as tags, on cloud resources and an agreed convention for servers make this repeatable. When attribution is not exact, a simple, stated rule is better than a perfect detail that has never been verified, for example splitting in proportion to use.

4. Look for the most common waste

Waste in a hybrid environment is almost always predictable. Resources created for a project and never deleted, machines oversized for real use, licences assigned to users who do not use them and services duplicated between public cloud and internal infrastructure are the main causes. The best method is to look at actual resource use over a sufficient period, not at theoretical configurations.

The basic rule is simple: if a resource has no owner and is not used, it should be proposed for decommissioning. The decision belongs to the service owner, not to whoever carried out the analysis. Those who analyse the data bring the evidence, those who approve decide, and the two roles should not be confused. Record the decision in the review minutes, otherwise the same resource will come back into discussion every time.

Type of wasteHow to recognise itTypical action
Orphaned resourcesNo associated owner and no recent accessCheck with the department, then decommission in a controlled way
Oversized machinesConsistently low use even at peak timesReduce the configuration after testing
Unused licencesAssigned to inactive users or decommissioned machinesRecover and reassign
Duplicate servicesSame data or same function across several environmentsChoose a single location with a migration plan

5. Govern with budgets and signals

Cost control works when thresholds and signals exist. A budget per service or per department, with alerts when it nears the limit, lets you act before the period ends. Thresholds should be set with the owners, not imposed from above, and reviewed when the business changes. A signal that arrives late, such as an unexpected invoice, is a failure of the process, not of the technology.

6. Make the review periodic

FinOps work does not end with an analysis. A periodic meeting, with finance, service owners and whoever runs the infrastructure, serves to review costs, anomalies and the decisions taken. The agenda is always the same: what has changed, what has been spent beyond expectations, which actions have been completed and which remain open. Every action has a named owner and a written deadline.

  1. Compare actual costs with the budget for each service.
  2. Analyse variances above the agreed threshold.
  3. Decide for each anomaly whether to accept it, correct it or investigate further.
  4. Record the decisions and assign owners and deadlines.

7. The balance point between cost and risk

Reducing costs must never remove a service's continuity or security. A backup deleted to save money, or redundancy removed from a critical system, can cost far more at the first incident. The right criterion is to compare the cost of a service with the damage its outage would cause, and invest accordingly. FinOps helps make this choice explicit, rather than leaving it to the rush of the last budget round.

8. How to start: a list of services with their costs

The first concrete step, to take this week, is a list of the main services with, for each, the owner, the estimated annual cost and the source of the figure. It need not be perfect: it is a starting point to be improved. Involve your accountant for depreciation rules and the provider for items you do not know. The provider can help you fill it in, but responsibility remains with the company.

Immediately afterwards, choose the two services with the highest or most uncertain cost and analyse, for each, the items, actual use and owner. That analysis gives rise to the first periodic review. Each service analysed should then be compared with the budget forecast, and the differences become the first item on the agenda. Do not look for savings straight away: first understand where the cost comes from.

Frequently Asked Questions

Is FinOps only for large companies with public cloud?

No. The method is useful whenever a cost is shared across several departments or systems, including a small company with in-house servers and a few cloud services. The process can be simple: a list of costs, an owner for each item and a monthly or quarterly review. Size matters less than clarity about who decides.

Do I need a specific tool to do FinOps?

It is not essential to start. A shared spreadsheet with the cost inventory, tags or naming conventions for resources and a review calendar is enough for the early stages. Tools become useful when volumes grow and manual collection no longer keeps up. Choose a tool only after defining the process, otherwise you will automate the chaos.

How do I compare the costs of in-house servers and public cloud?

The comparison is valid only if it includes every item: hardware, support, energy, space, licences, staff and exit costs. Comparing the cloud fee with the hardware invoice alone gives misleading results. Ask your adviser to build a total-cost comparison, with the assumptions stated, and to revisit it when workloads change.

How often should the cost analysis be reviewed?

It depends on how variable the costs are. For public cloud items that change often, a monthly review is advisable, while for stable contracts a quarterly review may be enough. In any case a full review should be carried out at least once a year, tied to development plans. What matters is that the review leads to recorded decisions.

Who should take part in the cost review?

At least three roles are needed: whoever approves the budget, whoever is responsible for the services and whoever runs the infrastructure. The participation of the departments that use the services is important, because they know the real need. An external provider can take part to bring data and comparisons, but the decisions remain with the company.

Can cutting costs increase risks?

Yes, if backups, redundancy, updates or monitoring are cut without assessing the effect. Every reduction should be compared with the damage that an outage or data loss would cause for that service. Safe reductions are usually those on orphaned and oversized resources, not on the protection of critical systems.

About this content

General informational guide. Regulations and technical requirements change over time: always check the official sources and assess your specific case with a qualified technical or legal advisor.

GET IN TOUCH

Talk to our team.

We'll show how Altovar integrates into your infrastructure. No commitment, no sales pitch. Just a technical conversation.

BOOK A DEMO →