Skip to content

Cost dashboard

The dashboard is the home screen and the main money view. It answers one question fast: how much is this cloud costing right now, and is anything unusual?

Everything on this page is scoped to the organisation you are currently working in and to the project and region filters you select at the top.

The hero: month-to-date cost

The large figure at the top is your month-to-date (MTD) cost, from the first of the current month until now.

Whenever possible this number is billed, not guessed: it comes from the OpenStack Rating API (CloudKitty), the same source your provider uses to invoice you. A badge tells you the source:

  • Billed (real): pulled from CloudKitty rating data.
  • Estimate: computed from a pricing grid (vCPU/hour, RAM GB/hour, disk GB/month, floating IP) when rating data is not available.

The hero breaks the total down into compute, storage and network.

ICU and currency

Provider costs are recorded internally in ICU (the provider's billing unit) and converted to your display currency on the fly. Switching currency never invalidates the data, so it is instant. The FAQ explains how ICU maps to CHF and EUR.

Forecast and trend

  • End-of-month forecast: a projection of where the current month will land if usage continues on its current path. It blends a recent-weighted average (EWMA) with a linear trend, and falls back to a simple linear projection when there is little history.
  • Month-over-month delta: how the current MTD compares with the same point in the previous month, shown as an amount and a percentage.
  • 30-day chart: a stacked area chart of daily cost by category (compute, storage, network) so you can see the shape of your spend, not just the total.

Top movers

Two lists surface what is driving change:

  • Top cost increases: what went up the most versus the previous period.
  • Top cost decreases: what went down.

Use these to explain a jump in the bill in seconds instead of scanning every resource.

Anomaly banner

When a recent day's cost is far outside the normal range for your organisation, a banner appears at the top of the page. Detection is statistical: OpenWatch keeps a rolling window of daily snapshots and flags a day whose cost is several standard deviations above the baseline (a z-score test), with a severity of mild or strong.

Tip

A downward swing (spend dropping) is treated as good news and stays silent. Only unexpected increases raise the banner.

Smaller widgets

  • Yesterday ticker: yesterday's total spend at a glance, so you notice a spike the morning after.
  • Budget widget: the hottest budgets by burn percentage. A budget that has consumed more than 80% of its allotment is highlighted as a warning. See Cost alerts for setting budgets and thresholds.
  • Savings opportunities: a summary of potential monthly savings, linking through to rightsizing recommendations.

Filters: projects, regions, currency

The controls at the top of the dashboard change what the whole page shows.

Control Options Effect
Project A single project, or all projects Focus on one project or aggregate the whole organisation
Region A single region (for example dc3-a, dc4-a), or all regions Focus on one region or aggregate across regions
Currency CHF or EUR Changes the display currency instantly

This produces four views:

  • Single project, single region: the detailed view of one project in one region.
  • Single project, all regions: one project aggregated across its regions.
  • All projects, single region: every project in one region.
  • All projects, all regions: the global organisation view.

Costs are global per project

Rating costs are the same regardless of which regional endpoint reports them, so OpenWatch counts them once per project to avoid double counting. Resource inventory, on the other hand, is per region.

The period selector

By default the dashboard shows MTD, backed by the billed rating figures. You can switch to other windows (for example a custom date range) using the period selector.

Billed vs rebuilt

MTD keeps the billed hero from the rating API. Other periods are rebuilt from OpenWatch's daily snapshots, which is instant but loses the "billed" tag for that window, because OpenWatch does not re-query the rating API for arbitrary ranges. If the snapshot history is empty, the view falls back to the MTD numbers.

Refreshing and the "preparing your data" state

The dashboard reads from a cache and never blocks on a live cloud call. This keeps the page fast even under load.

  • If the cache is fresh, you see current data immediately.
  • If it is stale, you see the most recent data with a small note that newer figures are being fetched in the background.
  • If it is empty (for example a project you just connected), you see a placeholder and a "preparing your cost data" banner. The page refreshes itself automatically after a few seconds once the background job has populated the cache.

Background refresh runs on a schedule, and you can also force it from the refresh action, which clears the cache and queues an immediate background fetch. Cost alerts are re-evaluated when you refresh manually.

Where to go next