Analyse

Tracking costs day to day

The cloud is billed by use: a forgotten resource, a test environment left running or an unexpected surge in load shows on the bill a month later. Tracking costs continuously means seeing the drift while there is still time to act.

Enlarge

Overview: the month's indicators, the spend of the last 12 months and the breakdowns.

Actual sizeClose

FinOps Move2Cloud reads your costs at each synchronisation, once a day by default, and shows you:

• the month's spend to date and the end-of-month forecast, compared with the previous month;
• twelve months of history, and the day-by-day record of the spend, to download as CSV;
• the breakdown by provider, by resource type, by region and by project or app;
• the savings to capture, priced when the data allows it.

All these views are exported as PDF or CSV, for accounting or a committee.

Understanding your figures

The tool shows three kinds of figures, which do not come from the same place: the spend observed, read at your providers; the end-of-month forecasts; the possible savings, estimated. The detail of each provider is in What is followed for each provider.

Spend observed

What your providers bill, as they give it, excluding tax, over twelve months: the current month up to the last synchronisation, then the eleven months before. One exception: Scalingo does not publish the current month's usage, which is estimated at the list price of what runs; its closed months come from its invoices, and a month just closed keeps the estimate until its invoice arrives.

• Credits and discounts: AWS is read without its credits or refunds; Google Cloud, credits deducted; Azure, at your contract's actual cost; Scalingo, closed months, with a “Discounts and adjustments” line that makes the total match the invoice.
• Currencies: the amounts billed in dollars (AWS, and Azure, Google Cloud or Scaleway when your account is billed in dollars) are converted to euros at the European Central Bank's reference rate of the day. At each synchronisation, the months read again are converted again at that same rate; the rate and its date are recalled under the overview's chart.
• A project visible from two connections is counted once, by the first one that followed it.

Forecasts

The end-of-month forecast comes from the provider when it publishes one: OVHcloud, AWS (Cost Explorer) and Azure (Cost Management). For Scalingo, the tool extends the price of what runs to the end of the month. For Scaleway and Google Cloud, which publish none, and for AWS or Azure when they refuse one (an account too recent, for example), it extends the month's spend at its current pace. A forecast is never shown below the spend already observed, and a closed month has none.

Savings estimated

The savings shown are estimates, to check before acting: the tool changes nothing at your providers.

• The tool's rules: for example, a non-production environment left on could be switched off outside working hours (Monday to Friday, 8 am to 6 pm, Paris time, that is 50 hours out of 168): the saving estimated is about 70% of its monthly cost, and is offered only above €10 a month.
• The provider's advice, when it publishes some: at AWS, Cost Optimization Hub's, with the saving AWS estimates.
• Commitments: the saving is priced with the discount you gave for OVHcloud, otherwise with a reference rate (OVHcloud: 15% over 12 months, 30% over 36 months; AWS: Savings Plans 30% over one year, 50% over three years, reserved instances 35% and 55%).
• Never counted twice: in a project asked to switch off outside working hours, the other advice on the same machines (resizing, stopping, committing) is another way to save; the totals count only the larger of the two.
• The AI's leads are marked as such and enter no total.

Freshness of the data

The figures date from the last synchronisation (once a day by default, depending on the plan). Providers often complete their amounts a little late: AWS gives the current day's costs partially, and several of its services only appear during the month; OVHcloud and Scaleway still adjust the previous month, which the tool reads again at each synchronisation; Google Cloud's export receives its first rows a few hours after it is turned on.

An example, from the invoice to the dashboard

An AWS account, on the 10th of the month. Cost Explorer gives $300 of usage cost since the 1st, and AWS's invoice will add $60 of taxes and take off $40 of credits. The tool takes the $300 of usage, without taxes or credits, and converts them at the day's ECB rate, for example $1 = €0.92: €276. AWS forecasts $650 more by the end of the month: the forecast shown is (300 + 650) × 0.92 = €874. Cost Optimization Hub advises resizing an instance, for $50 a month: the recommendation shows €46. The gap with AWS's invoice comes from the taxes, the credits and the exchange rate, never from something left out.

Why dashboards

Not everyone looks at the same figures: management wants the trend and the savings, a product team the cost of its applications, finance the gap to budget. A dashboard gives everyone the view that concerns them, without drowning them in the rest.

Enlarge

An open dashboard: its indicators, charts and lists, on its scope.

Actual sizeClose

In FinOps Move2Cloud, a dashboard is made of blocks (spend, forecast, breakdowns, recommendations, budgets, alerts) on a chosen scope: a provider, a cloud account, an environment, projects. It is created from a template, from a sentence the AI shapes, or from a blank page.

It stays private until its author shares it, then is shared with chosen people or the whole team, or through a public link valid for 24 hours. It thus becomes the natural support of a FinOps committee or a management review.

Why budgets and alerts

A budget is only worth something if it warns you before the bill. That is the role of alerts: to flag a drift while there is still time to correct it.

Enlarge

Budgets: the overall budget, the consumption and the forecast, then each scope with its budget and its state.

Actual sizeClose

In FinOps Move2Cloud, you set a global budget, a budget per cloud provider (all its projects and apps) and a budget per project or app, with an alert threshold. An e-mail is sent:

• when the spend reaches the chosen threshold;
• when the end-of-month forecast announces an overrun, before it happens;
• when the budget is exceeded.

Each alert is sent once per month and per budget, to inform without tiring. The recipients are chosen per organisation, and a weekly summary can be sent to the team.

Tagging and cost allocation

Allocating costs means being able to say who each euro belongs to: which product, which team, which environment, which customer. It is the condition of fair internal chargeback and of shared accountability.

Good tagging practices:

• a small number of keys, always the same: environment, application, team, cost centre;
• values written the same way everywhere (prod, preprod, staging…), decided once and documented;
• tags set when resources are created, by the deployment tooling rather than by hand;
• a regular check of the share of spending that escapes allocation.

FinOps Move2Cloud allocates your costs according to the structure your providers bill: Public Cloud projects at OVHcloud, apps at Scalingo, projects at Scaleway and Google Cloud, accounts at AWS, subscriptions at Azure. The environment of each project (production, non-production) is inferred from its name, and can be corrected by hand. Tags set on resources are not read at this stage: a clear split into projects and a consistent naming convention already give a reliable allocation, and dashboards can be filtered on project names.