{"canonical_url":"https://engineer.company/glossary/google-cloud/","content_markdown":"# Google Cloud\n\n## What it is\n\nGoogle Cloud is the part of Google that rents out computing from its own data\ncentres: virtual machines on Compute Engine, object storage in Cloud Storage,\nmanaged databases in Cloud SQL, Kubernetes clusters in GKE, the BigQuery data\nwarehouse and many more services, all paid for by use. Its older name, Google\nCloud Platform or GCP, is still in wide use, even in parts of Google's own\ndocumentation.\n\nEverything Google Cloud runs for a customer lives inside a project. A project is\nthe container for resources, for the APIs switched on for them, for the quotas\nthey draw on and for the link to the billing account that pays. It has an\neditable display name and a project ID that is unique across all of Google Cloud\nand permanent once the project exists. Projects can be grouped into folders,\nfolders can hold other folders, and the whole tree hangs from an organisation\nresource that belongs to a company's Google Workspace or Cloud Identity account.\nRoles and organisation policies set at the organisation or on a folder are\ninherited by every project beneath it, which makes a well-drawn tree easy to\ngovern and a badly drawn one hard to untangle.\n\nThe network is drawn differently from the other large clouds. A VPC network in\nGoogle Cloud, together with its routes and firewall rules, is a global resource\nthat belongs to no region; only its subnets are regional. A virtual machine in\nBelgium and another in Iowa can therefore share one private network and reach\neach other on internal addresses, with no VPN or peering between the two\nregions. On AWS, by contrast, each VPC lives in a single region.\n\nIdentity comes in two kinds. People sign in with Google accounts and receive IAM\nroles on a project, a folder or the whole organisation. Programs use service\naccounts: identities that belong to one project and that a virtual machine, a\ncontainer or a build job can act as. Attach a service account to a Compute\nEngine instance, and the code running there can call Google APIs with that\naccount's roles without a password or key file on the disk.\n\n## Why it matters\n\nFor many teams the reason to pick Google Cloud is data. BigQuery is a warehouse\nwith no servers to size: tables sit in datasets inside a project, queries are\nwritten in standard SQL, and on the on-demand model each query is billed by the\nbytes it reads, while capacity pricing pays for a reserved amount of processing,\nmeasured in slots, instead. Storage is columnar, so a query pays for the columns\nit touches rather than for whole rows. Cloud Storage sits beside it as the\nlanding zone, and loading files from a bucket is a standard way in. A team that\nwould elsewhere provision and tune a warehouse cluster can start with a bucket,\na dataset and a query.\n\nThe project boundary is the second reason, because so much attaches to it.\nBilling, quotas, enabled APIs and IAM all meet at the project, so one project\nper application and environment draws a clean line: production and test get\nseparate permissions and separate lines on the bill. Shutting a project down\nstops its billing and every resource in it at once. Its owners can restore it\nfor 30 days, though some services, Cloud Storage among them, may delete their\ndata sooner. Above the projects, an organisation policy set once on a folder can\nforbid whole classes of risky setting, such as default networks or downloadable\nservice account keys, in every project beneath it. That is how a small team\nkeeps a dozen projects consistent without checking each one by hand.\n\n## What it looks like in practice\n\nThree portfolio entries show Google Cloud as the landing zone for a data\nwarehouse, the offsite store for backups and one end of a private network.\n\nIn the data work,\n[six ETL/ELT pipelines fed BigQuery from a warehouse behind an IPsec VPN](/portfolio/designed-and-ran-6-etl-elt-pipelines/).\nCurated datasets and tables were exported, compressed with gzip and transferred\nto a Google Cloud Storage bucket, and the upload and schema mapping into\nBigQuery were automated with the bq CLI and Python ingestion scripts. Once the\nCloud Storage hop and the BigQuery load ran on their own, the whole job took\nunder 20 minutes instead of 3+ hours by hand, BigQuery was refreshed daily\nrather than weekly, and the Python APIs reading from it gained 30% in\nperformance.\n\nFor backups,\n[a hundred critical data backups were automated](/portfolio/automated-100-critical-data-backups-using-barman/),\nwith Barman handling the PostgreSQL side and Google Cloud holding the offsite\ncopies. Bash and Python orchestrated and validated the whole regime, with\nretention policies to keep the copies current and integrity checks to confirm\nthat each one was good rather than merely present. The backups then ran\nautomatically and could be verified across every database.\n\nOn the network side,\n[eight WireGuard and IPsec tunnels between Google Cloud and on-premises Linux systems](/portfolio/managed-8-wireguard-and-ipsec-vpn-connections/)\nwere managed and troubleshot as one set. Their configuration was standardised,\ntheir health was monitored, and the recurring routing and key-exchange problems\nwere fixed at the root rather than with a restart, after which all eight tunnels\nran securely and reliably.\n\n## Where it goes wrong\n\nGoogle Cloud's sharpest edges are its conveniences: settings that make the first\nafternoon easy and are still in place a year later, long after anyone remembers\nchoosing them.\n\n**The default network stays open.** Unless an organisation policy prevents it,\nevery new project starts with a default network that has a subnet in every\nregion and four pre-populated firewall rules. Three of them accept traffic from\n0.0.0.0/0, which is the whole internet: SSH on tcp:22, Remote Desktop on\ntcp:3389, and ICMP. The fourth allows all TCP, UDP and ICMP traffic from\n10.128.0.0/9 inside the network. Any virtual machine given an external address\non that network is therefore open to SSH attempts from anywhere. The\n`compute.skipDefaultNetworkCreation` constraint keeps new projects from getting\none, and a custom network with rules that name their sources replaces it.\n\n**A service account key ends up in a repository.** A user-managed key is a JSON\nfile that lets whoever holds it act as the service account, and by default it\nstays valid until someone deletes it. Such files spread easily: into CI secrets,\ninto container images and, by mistake, into repositories. Google's guidance is\nto choose a safer alternative wherever one exists. Workloads on Google Cloud use\nthe service account attached to the resource they run on, and workloads\nelsewhere, such as a CI system or another cloud, use Workload Identity\nFederation, which trades the outside system's own credentials for short-lived\nGoogle tokens. Organisations created on or after 3 May 2024 have key creation\nand key upload disabled by default through organisation policy constraints, and\nolder ones can set the same constraints.\n\n**A small query reads a large table.** On the on-demand model BigQuery bills for\nthe bytes a query reads, and on a table that is not clustered a `LIMIT` clause\nleaves that number unchanged: `SELECT * ... LIMIT 10` returns ten rows and bills\nevery byte of every column. Select only the columns needed, partition large\ntables by date so that queries filtering on it skip old partitions, check the\nestimate from a dry run before running anything new, and set maximum bytes\nbilled. A query whose estimate exceeds that limit fails before it runs, without\nincurring a charge.\n\n**Cold storage holds warm data.** Cloud Storage offers Nearline, Coldline and\nArchive classes that cost less per gigabyte stored than Standard, in exchange\nfor retrieval fees and minimum storage durations of 30, 90 and 365 days. An\nobject deleted, replaced or moved before its minimum is charged as if it had\nbeen stored for the whole period. Backups that rotate daily, or scratch files a\npipeline writes and removes within the hour, can therefore cost more in Archive\nthan they would in Standard. The colder classes suit data that is written once\nand read rarely, and a lifecycle rule can move objects into them as they age.\n\n## Further reading\n\n- [Google Cloud Documentation](https://cloud.google.com/docs) — Google's index\n  of guides, references and tutorials for every Google Cloud product.\n- [About resource hierarchy](https://cloud.google.com/resource-manager/docs/cloud-platform-resource-hierarchy)\n  — how organisations, folders and projects nest, and how roles and policies\n  flow down them.\n- [VPC firewall rules](https://cloud.google.com/firewall/docs/firewalls) — how\n  firewall rules are evaluated, including the rules the default network starts\n  with.\n- [Best practices for managing service account keys](https://cloud.google.com/iam/docs/best-practices-for-managing-service-account-keys)\n  — how keys leak, how to restrict them and what to use instead.\n- [Estimate and control costs](https://cloud.google.com/bigquery/docs/best-practices-costs)\n  — BigQuery's guidance on dry runs, maximum bytes billed and why `LIMIT` saves\n  little.\n- [Storage classes](https://cloud.google.com/storage/docs/storage-classes) — the\n  classes side by side, with their minimum durations, retrieval fees and\n  availability.\n\n\u003chttps://engineer.company/glossary/google-cloud/\u003e\n","date_modified":"2026-10-21T06:22:55Z","description":"How Google rents out computing through projects, folders and one global network, and where Google Cloud estates leak money, keys and open ports.","formats":{"gemtext":"https://engineer.company/glossary/google-cloud/index.gmi","gopher":"https://engineer.company/glossary/google-cloud/gophermap","html":"https://engineer.company/glossary/google-cloud/","json":"https://engineer.company/glossary/google-cloud/index.json","markdown":"https://engineer.company/glossary/google-cloud/index.md","pdf":"https://engineer.company/glossary/google-cloud/index.pdf"},"kind":"page","language":"en","title":"What is Google Cloud?","url":"https://engineer.company/glossary/google-cloud/","version":1}