This chapter explains how Google Cloud billing works, service by service: the SKU Google meters, the price that SKU carries in Frankfurt, and the row it becomes in the billing export to BigQuery and the FOCUS export. Reading a Google Cloud bill takes three things per service: the pricing dimensions, the SKUs under which the usage appears, and the levers that change the amount. Every page of this chapter supplies exactly those three for one service or one group of services, with list prices in USD for europe-west3 read from Google's pricing pages on the date each page names.

What this chapter covers

The chapters before this one explained how a cloud bill comes about, how billing data is structured, and what cloud services cost at AWS, Azure and Google Cloud side by side. This chapter stays with one provider and goes deep. Google bills every service through SKUs that name the resource, the region and the kind of usage: a VM is cores and RAM per family, each its own SKU, plus disk, licence and traffic; a bucket is bytes per class and location plus operations per 1,000; a warehouse is bytes scanned or slot-hours; a model is tokens in and tokens out. One architecture therefore produces dozens of SKUs, and the discounts never change the price in the row but arrive as credits beside it, sustained use, committed use, free tier and promotions each with their own type. The pages name those SKUs and credit types so a row in the export can be read back to the resource and the decision that caused it.

Seven sections structure the chapter:

Every service page follows the same structure: the pricing dimensions as a table, the SKUs and columns under which the usage appears in the export and in FOCUS, the levers by configuration, purchase option and architecture, the ways it goes wrong with the symptom in billing data, one example with list prices in USD for Frankfurt, and the sources with their read date. The billing-mechanics pages replace the export section with invoices, exports and APIs, because they describe no metered service but what shapes the invoice. Frankfurt prices were read from Google's pricing pages with the region selected, and where a service prices the same everywhere, the page says so.

What this chapter is not: a comparison across providers and not a tool manual. The comparison is the chapter What cloud services cost; the console's reports and exports are the chapter Tooling, with Google Cloud billing reports as the page for Google's own tool. A price on these pages is a list price in USD read on the stated date; an invoiced account, a reseller or a negotiated agreement changes it, and the export's price.effective_price is what was actually paid.

Key terms

  • SKU: the unit Google meters and prices, named by service, resource, kind and region; every page lists the SKUs of its service.
  • line item: one row of the billing export, one SKU of one project in one hour, with usage, cost and credits.
  • usage export: the billing export to BigQuery, standard, detailed or pricing, the source every page reads its columns from.
  • committed use discount: Google's commitment to resources or spend over one or three years, returned as a credit beside the usage it covers.
  • credit: the negative amount beside a row that carries every discount, allowance and promotion with its type.
  • billing account: the account that pays for a set of projects and decides the scope of budgets, exports and commitments.
  • tag: on Google Cloud both the label on a resource and the tag bound to the hierarchy, which the export carries apart.
  • effective cost: the cost after credits and with commitments amortised, the figure a chargeback uses rather than the list amount.

Neighbouring chapters

Before this chapter stand AWS billing and pricing and Azure billing and pricing, the same structure for the other two providers, and before them What cloud services cost, whose cross-provider pages this chapter's pages are children of: virtual machines and instance families, managed databases, block, object and file storage, data transfer and egress and AI and machine learning services. After it comes Kubernetes cost, which takes the cluster this chapter prices as a node pool and a pod request and follows it into the workload.

Two chapters supply what these pages assume. Billing data explains the export every page names its columns from: the Google Cloud billing export and FOCUS columns. Pricing models explains the purchase options the levers point to: Google Cloud committed use discounts, sustained use and tiered discounts and spot and preemptible capacity. The full list of chapters is in the handbook.