{"canonical_url":"https://engineer.company/glossary/aws/","content_markdown":"# AWS\n\n## What it is\n\nAWS, short for Amazon Web Services, is Amazon's business of renting out\ncomputing by the hour and the gigabyte: virtual servers, file storage, managed\ndatabases, private networks, queues and a long catalogue of further services,\nall switched on through one API and billed to one account. Amazon has sold it\nsince 2006. Two of its best-known services explain much of the rest. Amazon EC2\nrents virtual machines, which AWS calls instances, and Amazon S3 stores files as\nobjects inside containers called buckets.\n\nIn the vocabulary of the NIST definition, AWS is a public cloud: anyone with a\npayment card can open an account, resources appear within minutes, and every\nsecond or gigabyte of use is metered. It is usually weighed against Microsoft's\nAzure and Google Cloud, which sell much the same building blocks under other\nnames. What AWS calls an EC2 instance is an Azure virtual machine or a Compute\nEngine instance on Google Cloud.\n\nEvery AWS resource lives somewhere specific. The service is divided into\nRegions, each designed to be isolated from the others, and every Region holds\nseveral Availability Zones. A zone is one or more discrete data centres with\nredundant power, networking and connectivity, physically separate from the other\nzones in its Region. A resource is created in one Region, often in a single\nzone, and a team chooses the place with latency, law and price in mind.\n\nAccess to all of it runs through AWS Identity and Access Management, or IAM. An\naccount starts with a root user that can do anything, and IAM adds users,\ngroups, roles and policies that say which API call is allowed on which resource\nunder which conditions. Most questions about AWS security turn out to be\nquestions about those policies.\n\n## Why it matters\n\nTeams pick AWS for its range. A product that needs a managed PostgreSQL, a\nqueue, a content delivery network, object storage and somewhere to run\ncontainers can have all of them under one bill and one identity system,\ndocumented to a common pattern. The range is also the price. Each service has\nits own pricing dimensions, its own limits and its own corner of the IAM policy\nlanguage, and a team that adopts ten services now maintains ten small\nintegrations with Amazon's interfaces. Leaving later means replacing each one.\n\nThe second decision is what stays with the customer. AWS publishes a shared\nresponsibility model and draws the line in two phrases: AWS is responsible for\nsecurity “of” the cloud, meaning “the hardware, software, networking, and\nfacilities that run AWS Cloud services”, and the customer for security “in” the\ncloud. Where the line falls depends on the service. On an EC2 instance the\ncustomer patches the guest operating system, looks after every program installed\non it and configures the security group, the firewall in front of the instance.\nOn S3 or DynamoDB, AWS runs the operating system and the platform, and the\ncustomer is left with the data, its encryption options and the permissions that\ndecide who may read it.\n\nThe third is cost, which on AWS is a property of the design more than an invoice\nthat arrives later. Amazon's Well-Architected Framework names six pillars to\nreview a workload against: operational excellence, security, reliability,\nperformance efficiency, cost optimization and sustainability. Cost earns its\nplace on that list because the same system can be built cheaply or expensively\nby choices made in its first week: instance sizes, how many zones, how much data\nleaves the Region, and which managed services stand in for a process on a plain\nserver.\n\n## What it looks like in practice\n\nOn [the tech stack page](/tech-stack/), AWS sits beside Google Cloud as one of\nthe two clouds the company works in by default, with Azure added where a client\nalready runs there.\n\nOne engagement used both of those clouds at once:\n[twenty Docker applications ran across Google Cloud and AWS](/portfolio/ran-20-docker-apps-on-google-cloud-and-aws/),\nsome of them R-based analytics apps. They were deployed and maintained with the\nsame configuration and monitoring in each cloud, so that no app turned into a\nspecial case of its own. When something broke, troubleshooting went through\nPodman and orchestration debugging through Kubernetes, and the analytics apps\nwere kept stable and reproducible in production on both sides.\n\nAnother engagement ran the other way. For an international client,\n[an AWS setup had been over-provisioned and had collected unused services](/portfolio/a-tenth-of-the-cloud-bill-and-nothing-lost/),\nand the cost had grown out of proportion to the business's needs. The work reviewed which\nservices were used, removed unnecessary services, adjusted the remaining capacity and moved\nworkloads away from AWS. Those infrastructure changes are the delivered work described by the\nrecord. It does not establish a like-for-like cost ratio, a productivity measurement or a\ndisruption-free migration. The example is useful for its method of examining the existing\nsetup and changing what the business needs to run, not as a promise of a particular saving.\n\nBoth cases come down to the same habit. A pay-per-use service bills for whatever\nwas switched on, so an AWS estate stays healthy when someone keeps checking what\nis running against what the business uses.\n\n## Where it goes wrong\n\nAWS rarely loses data or stays down for long. The expensive incidents tend to\nstart inside the customer's own account.\n\n**The root user does the day's work.** Every account begins with a root user,\nthe email address and password that opened it, with unrestricted access to every\nresource and setting, including closing the account. Signing in as root for\nroutine tasks means that one leaked password reaches everything. AWS's guidance\nis to safeguard the root credentials, put multi-factor authentication on them\nand keep them for the handful of tasks that need them, while everyday work\nhappens through federated sign-in and roles.\n\n**Long-lived access keys leak.** An access key pair works until someone deletes\nit, and it travels well: into a laptop's dotfiles, a CI variable, a container\nimage, a commit. The IAM best-practice list starts from the other end. People\nsign in through an identity provider and receive temporary credentials, and\nworkloads take temporary credentials from an IAM role attached to the compute\nthey run on. Where a long-term key is truly required, rotate it, and remove\nkeys, users and roles that have gone unused.\n\n**A budget alert is taken for a brake.** AWS Budgets sends a notification when\nactual or forecasted spend crosses a threshold, and its figures update up to\nthree times a day, so an alert reports money that is already spent. The\nresources keep running and keep billing. Budgets can also take an action, such\nas applying an IAM policy or a service control policy that prevents new\nresources from being created, or acting on specific EC2 or RDS instances. It\ndoes so only when that action was configured in advance, with a role to act\nthrough.\n\n**Everything sits in one Availability Zone.** A single instance with a single\ndisk in one zone is the cheapest layout, and it shares the fate of that one\nlocation. Zones exist so that a power or network fault in one leaves the others\nrunning, and the AWS documentation recommends spreading an application across\nseveral. A database that matters gets a standby in a second zone, and a web tier\ngets instances in at least two, behind a load balancer that spans both.\n\n**A bucket opened for a demo stays open.** New S3 buckets block public access by\ndefault, but a bucket policy or an object permission can change that in a\nminute, and a link shared with a client for one afternoon is still public a year\nlater. S3 Block Public Access, set for the whole account or organisation,\noverrides those policies. Turn it on wherever an account has no reason to\npublish files, and serve public files from a bucket built for that purpose.\n\n## Further reading\n\n- [AWS Documentation](https://docs.aws.amazon.com/) — the index of user guides,\n  API references and tutorials for every AWS service.\n- [Shared responsibility](https://docs.aws.amazon.com/wellarchitected/latest/security-pillar/shared-responsibility.html)\n  — the Well-Architected security pillar's account of which layers AWS secures\n  and which the customer does.\n- [The pillars of the framework](https://docs.aws.amazon.com/wellarchitected/latest/framework/the-pillars-of-the-framework.html)\n  — the six areas AWS asks a workload to be reviewed against, cost and\n  reliability among them.\n- [AWS Regions and Availability Zones](https://docs.aws.amazon.com/global-infrastructure/latest/regions/aws-regions-availability-zones.html)\n  — how Regions and zones are isolated, and why an application should span\n  several zones.\n- [Security best practices in IAM](https://docs.aws.amazon.com/IAM/latest/UserGuide/best-practices.html)\n  — federation, roles, multi-factor authentication, the root user and least\n  privilege, in one list.\n- [Managing your costs with AWS Budgets](https://docs.aws.amazon.com/cost-management/latest/userguide/budgets-managing-costs.html)\n  — budgets on actual and forecasted spend, their notifications and the actions\n  they can run.\n\n\u003chttps://engineer.company/glossary/aws/\u003e\n","date_modified":"2026-11-08T02:22:23Z","description":"What Amazon Web Services rents out, who answers for which part of its security, and the accounts, keys, budgets and buckets where AWS estates go wrong.","formats":{"gemtext":"https://engineer.company/glossary/aws/index.gmi","gopher":"https://engineer.company/glossary/aws/gophermap","html":"https://engineer.company/glossary/aws/","json":"https://engineer.company/glossary/aws/index.json","markdown":"https://engineer.company/glossary/aws/index.md","pdf":"https://engineer.company/glossary/aws/index.pdf"},"kind":"page","language":"en","title":"What is AWS?","url":"https://engineer.company/glossary/aws/","version":1}