Cloud Computing & DevOps Vocabulary: 100 Key English Terms
Cloud and DevOps teams around the world work in English: provider consoles, documentation, pipeline logs, alerts and incident channels all share the same vocabulary. Knowing these terms helps engineers read error messages and runbooks quickly, follow on-call handovers and postmortems, and explain infrastructure decisions clearly to colleagues in international teams.
100 terms
11 topics
55 core terms
201 example sentences
Read, understand, keep going
No account needed. Read the meaning, see where the term is used, then study the examples. Click any word in the lesson for help. You can also select a word and press Alt + Enter.
Cloud Basics and Service Models
cloud computing
noun
Meaning
Renting servers, storage, databases and other computing resources over the internet from a provider, instead of buying and running your own hardware.
Where you use it
The umbrella term for the whole field. In daily talk people simply say 'the cloud': 'we moved to the cloud', 'it runs in the cloud'. The formal 'cloud computing' appears in job titles, courses, reports and contracts.
Examples
Cloud computing lets a small startup launch in a week without buying a single server.
Our company started its move to cloud computing three years ago, and the old data center closes next month.
Common phrases
move to the cloud · cloud computing platform · cloud migration · in the cloud · cloud-based service
A company that owns large data centers and rents computing, storage, networking and other services to customers over the internet.
Where you use it
The three largest are Amazon Web Services (AWS), Microsoft Azure and Google Cloud, sometimes called hyperscalers. Teams talk about choosing, switching and depending on a provider; 'vendor lock-in' is the worry of being stuck with one.
Examples
Our cloud provider had a problem in one region, and three of our services went down.
Before you pick a cloud provider, compare the prices of the services you'll actually use.
Common phrases
major cloud providers · switch cloud providers · vendor lock-in · public cloud provider
A geographic area where a cloud provider has a group of data centers, such as Frankfurt or Ireland. You choose the region where your resources run.
Where you use it
Each region has a code, such as eu-central-1 (AWS, Frankfurt). Teams choose a region for lower latency to users, lower price or to keep data in a certain country for legal reasons. Each region contains several availability zones.
Examples
We deploy to two regions, so a full outage in Frankfurt won't take us offline.
The bucket was created in the wrong region, which is why downloads were so slow.
Common phrases
choose a region · multi-region · region outage · deploy to a region · cross-region replication
Short for infrastructure as a service: renting basic building blocks such as virtual machines, disks and networks, while you manage the operating system and everything above it.
Where you use it
Usually compared with PaaS and SaaS to show how much the provider manages for you. Amazon EC2, Azure Virtual Machines and Google Compute Engine are typical IaaS products. Often said 'eye-az'.
Examples
With IaaS, the provider looks after the hardware, but patching the operating system is still our job.
We started on IaaS because we needed full control over the servers.
Common phrases
IaaS provider · IaaS offering · IaaS model · move to IaaS
Short for platform as a service: a ready-made platform where you only upload your code, and the provider runs the servers, operating system and runtime for you.
Where you use it
Pronounced like the word 'pass'. Heroku, Google App Engine, Azure App Service and AWS Elastic Beanstalk are common examples. PaaS saves operations work but gives you less control than IaaS.
Examples
As a two-person team, we picked a PaaS so we wouldn't have to manage servers at all.
The PaaS handles scaling, but it doesn't let us install custom system packages.
Common phrases
PaaS provider · PaaS platform · deploy to a PaaS · PaaS offering
Short for software as a service: complete software that customers use through a browser or app and pay for by subscription, while the vendor runs everything behind it.
Where you use it
Pronounced 'sass'. Email services, CRM systems and team chat tools are typical SaaS products. It is also used as an adjective for companies: 'a SaaS startup', 'SaaS business model'.
Examples
We replaced our self-hosted ticket system with a SaaS tool, so there's one less server to maintain.
Describes a way of running code in which the provider starts and scales servers automatically for each request, and you pay only for the time your code actually runs.
Where you use it
Used as an adjective ('serverless function', 'serverless architecture') and as a noun ('we use serverless'). AWS Lambda and Azure Functions are well-known examples. A 'cold start' is the short delay when a function starts after being idle.
Examples
The image resizing runs as a serverless function, so it costs almost nothing when nobody uploads.
Serverless is great for spiky traffic, but cold starts can hurt latency.
Usage note
Servers still exist; 'serverless' only means you don't manage them yourself.
Common phrases
serverless function · serverless architecture · go serverless · cold start
Describes servers and software that a company runs in its own building or data center, rather than renting them from a cloud provider.
Where you use it
Often shortened to 'on-prem' in speech. Used before nouns ('on-premises servers') and after verbs ('run on-premises'). Banks and public institutions often keep some systems on-premises for regulatory reasons.
Examples
Our customer data stays on-premises, but the website runs in the cloud.
Moving from on-premises hardware to the cloud took us almost two years.
Usage note
Many people say 'on-premise', but a premise is an idea; premises (with -s) means a building. 'On-premises' or 'on-prem' is the correct form.
Common phrases
on-premises servers · run on-premises · on-premises data center · on-premises to cloud migration
A setup that combines a company's own on-premises systems or private cloud with a public cloud provider, with the two connected and working together.
Where you use it
Common in banks, telecoms and large enterprises that cannot move everything to the cloud. Compare 'public cloud' (shared provider), 'private cloud' (for one organization) and 'multi-cloud' (several public providers).
Examples
We use a hybrid cloud: core banking stays in our data center, and the mobile backend runs on a public cloud.
A hybrid cloud needs a secure, fast connection between the two sides.
Usage note
Hybrid cloud mixes your own infrastructure with a public cloud; multi-cloud means using more than one public cloud provider.
One of several separate data center locations inside a cloud region, each with its own power and network, so a failure in one should not affect the others.
Where you use it
Usually shortened to AZ. Engineers spread servers 'across multiple AZs' to survive a data center failure. AWS and Azure use 'availability zone'; Google Cloud simply says 'zone'.
Examples
The database has a standby copy in another availability zone, so it survived the power failure.
All our nodes were in a single availability zone, which was a big mistake.
Common phrases
across availability zones · multi-AZ · single availability zone · zone failure
An application, service or job that runs on cloud resources and uses computing power, for example a website, a batch job or a database.
Where you use it
A general word in cloud planning and documentation: 'migrate workloads to the cloud', 'production workloads', 'stateless workloads'. In Kubernetes, Deployments and Jobs are called workload resources.
Examples
We moved the batch workloads to spot instances, which cut their cost by more than half.
Not every workload belongs in the cloud; some legacy systems are cheaper to run on-premises.
Usage note
In everyday English 'workload' means the amount of work a person has. In the cloud it means a running application or job.
Common phrases
production workload · migrate workloads · run a workload · stateless workload
A cloud service where the provider handles setup, patching, backups and scaling, so you only use it and do not manage the servers underneath.
Where you use it
Common examples are managed databases (Amazon RDS, Azure SQL Database, Cloud SQL), managed Kubernetes and managed queues. Teams choose a managed service to reduce operations work, at a higher price per unit.
Examples
We switched to a managed service for PostgreSQL, so backups and failover are no longer our problem.
A managed service costs more per hour, but it saves us a full-time engineer.
Common phrases
fully managed service · managed database · managed Kubernetes · use a managed service
A computer, physical or virtual, that runs services and answers requests from other computers over a network, for example by sending web pages or data.
Where you use it
Used for both the machine and the software that serves requests: 'the web server', 'the database server', 'Nginx is our web server'. In the cloud, a server is usually a virtual machine you rent rather than a box you own.
Examples
The server ran out of disk space at 3 a.m., so the API stopped writing logs.
We replaced the three old physical servers with a small Kubernetes cluster.
Common phrases
web server · database server · restart a server · server load · physical server
A complete computer, with its own operating system, that runs as software on a physical machine. One physical server can host many virtual machines at once.
Where you use it
Usually shortened to VM, said 'V-M'. You create, start, stop, resize and delete VMs. On AWS a VM is called an EC2 instance; Azure and Google Cloud use the term virtual machine or VM directly.
Examples
I created a small virtual machine to test the new Linux version before we upgrade.
Unlike a container, a virtual machine needs a few minutes to boot its full operating system.
Common phrases
create a virtual machine · VM size · spin up a VM · virtual machine image · guest operating system
One running copy of a virtual server or service in the cloud. You choose its size, such as how many CPUs and how much memory it has, and pay while it runs.
Where you use it
Very common on AWS (EC2 instance, RDS instance) and in general talk: 'launch an instance', 'stop the instance', 'instance type'. It can also mean one copy of an application: 'we run three instances of the API'.
Examples
Please stop the test instances over the weekend so we don't pay for them.
We moved the database to a larger instance type with 64 GB of memory.
Usage note
In everyday English 'for instance' means 'for example', but in the cloud an instance is a running server, not an example.
Common phrases
launch an instance · instance type · stop an instance · EC2 instance · terminate an instance
To quickly create and start a new server, container or environment, usually with a command or a few clicks.
Where you use it
Informal but used everywhere in the industry: 'spin up an instance', 'spin up a test environment'. The past form is 'spun up'. The opposite is 'spin down' or 'tear down'.
Examples
Can you spin up a copy of staging so I can test the migration?
The CI job spins up a database container, runs the tests and then removes it.
Common phrases
spin up an instance · spin up a cluster · spin up a test environment · spin up containers
Software that creates and runs virtual machines, sharing one physical computer's CPU, memory and storage between them and keeping them separate.
Where you use it
Type 1 hypervisors run directly on the hardware (VMware ESXi, Microsoft Hyper-V, KVM, Xen); type 2 hypervisors run on top of a normal operating system (VirtualBox). Cloud users rarely touch the hypervisor directly.
Examples
Our on-premises hypervisor hosts about forty virtual machines on four physical servers.
The cloud provider patched the hypervisor, and some instances had to be restarted.
To delete a server, environment or set of resources completely when it is no longer needed, usually with a command or script.
Where you use it
The opposite of 'spin up'. The past form is 'tore down', the past participle 'torn down', and the noun is 'teardown'. With Terraform, 'terraform destroy' tears down everything in a configuration.
Examples
The pipeline tears down the preview environment as soon as the pull request is merged.
Remember to tear down the load test cluster tonight, or it will keep costing us money.
Usage note
'Tear' here rhymes with 'air' (to pull apart), not with 'ear' (a drop from your eye).
Common phrases
tear down an environment · tear down a cluster · spin up and tear down · teardown script
A lightweight, isolated package that holds an application together with everything it needs to run, so it behaves the same on a laptop, a test server or in the cloud.
Where you use it
Containers share the host's operating system kernel, so they start in seconds and use less memory than virtual machines. You build, run, stop and restart containers; Docker is the best-known tool.
Examples
The container keeps restarting because the app crashes when the config file is missing.
Each microservice runs in its own container, so we can update them independently.
Usage note
In everyday English a container is any box or jar. In this field it always means a software container unless people are talking about shipping.
Common phrases
run a container · container image · container runtime · restart a container · containerized app
A widely used platform for building container images and running containers. It made containers popular after its release in 2013.
Where you use it
A product name, so it is written with a capital D, but in speech it is also used as a verb or adjective: 'dockerize the app', 'Docker image', 'Docker Compose file'. Many people use 'Docker' to mean containers in general.
Examples
Install Docker on your laptop and run the whole stack with one command.
We dockerized the legacy app so it can finally run on Kubernetes.
An open-source system that runs and manages containers across many machines: it starts them, restarts failed ones, spreads traffic and scales them up or down.
Where you use it
Originally designed at Google and released in 2014, now maintained by the Cloud Native Computing Foundation. Often written K8s (eight letters between K and s). Managed versions include Amazon EKS, Azure AKS and Google GKE.
Examples
Kubernetes restarted the crashed pods automatically, so users didn't notice anything.
Running Kubernetes yourself is a lot of work, so we switched to a managed cluster.
Usage note
The name comes from the Greek word for helmsman or pilot. Stress the third syllable: koo-ber-NET-eez.
Common phrases
Kubernetes cluster · deploy to Kubernetes · managed Kubernetes · Kubernetes manifest · run on Kubernetes
A read-only package with an application, its libraries and settings, used as the template from which containers are started.
Where you use it
You build an image, tag it with a version, push it to a registry and pull it on servers. People often just say 'image' or 'Docker image'. Small images download and start faster.
Examples
The new container image is 300 MB smaller because we switched to a slim base image.
Always tag the container image with the commit hash, not just 'latest'.
Common phrases
build an image · push an image · pull an image · base image · image tag
A group of connected machines that work together as one system, for example to run containers in Kubernetes or to store data in a database.
Where you use it
Most often heard as 'Kubernetes cluster', but also 'database cluster' and 'Elasticsearch cluster'. You create, upgrade, scale and delete clusters. Teams often keep separate clusters for staging and production.
Examples
We're upgrading the production cluster this weekend, so expect a few pod restarts.
The cluster has six nodes spread across three availability zones.
Common phrases
Kubernetes cluster · create a cluster · upgrade the cluster · cluster autoscaler · database cluster
One machine, physical or virtual, that is part of a cluster. In Kubernetes, worker nodes are the machines where pods actually run.
Where you use it
Nodes are added, drained (emptied of pods before maintenance), cordoned and removed. A 'node pool' or 'node group' is a set of nodes with the same machine type. Do not confuse it with Node.js.
Examples
One node ran out of memory, and Kubernetes moved its pods to the others.
Drain the node before you patch it, otherwise users will see errors.
Common phrases
worker node · node pool · drain a node · add a node · node failure
The smallest unit that Kubernetes runs: one or more containers that are always placed together on the same node and share a network address and storage.
Where you use it
Pods are created, scheduled, restarted, evicted and killed. Engineers check pod status with kubectl: Running, Pending or CrashLoopBackOff. Pods are temporary; a Deployment replaces them when they die.
Examples
The pod is stuck in CrashLoopBackOff, so let's look at its logs.
Each pod runs the app container plus a small sidecar that ships logs.
Common phrases
pod status · restart a pod · pod crashed · pending pod · kill a pod
The automatic management of many containers or services: deciding where they run, starting and restarting them, connecting them and scaling them.
Where you use it
Usually in the phrase 'container orchestration'. Kubernetes is the dominant orchestration tool; others include Docker Swarm, HashiCorp Nomad and Amazon ECS. The tool itself is called an orchestrator.
Examples
Once we had forty containers, we needed proper orchestration instead of shell scripts.
Container orchestration takes care of restarts, so you don't have to watch every service.
Usage note
Stress is on 'stra': or-ke-STRAY-shun. The 'ch' is pronounced like 'k'.
A text file with step-by-step instructions, such as FROM, COPY and RUN, that Docker follows to build a container image.
Where you use it
Usually named exactly 'Dockerfile', with a capital D and no extension, in the project's root folder. Engineers write, review and optimize Dockerfiles; multi-stage builds keep final images small.
Examples
Add the new dependency to the Dockerfile, or the image won't build in CI.
I rewrote the Dockerfile as a multi-stage build, and the image went from 1 GB to 150 MB.
Common phrases
write a Dockerfile · multi-stage Dockerfile · Dockerfile instruction · base image
A storage service for container images, where pipelines push new images and servers or clusters pull them when they need to run them.
Where you use it
Often just 'the registry'. Examples are Docker Hub, Amazon ECR, Azure Container Registry, Google Artifact Registry and GitHub Container Registry. Many registries scan images for known vulnerabilities.
Examples
The pods can't start because the cluster has no permission to pull from the container registry.
Our private container registry deletes untagged images after two weeks.
Common phrases
push to the registry · pull from the registry · private container registry · image scanning
A way of working in which developers and operations engineers share responsibility for building, releasing and running software, using automation to deliver changes quickly and safely.
Where you use it
Formed from 'development' and 'operations' and widely used since around 2009. It names a culture and set of practices, but in job ads it is also a role: DevOps engineer, DevOps team. Some people say DevOps should not be a separate team, only a way of working.
Examples
After we adopted DevOps practices, we went from one release a month to several a day.
We're hiring a DevOps engineer with Kubernetes and Terraform experience.
To put a new version of an application or service onto servers so that it starts running there, for example in a test environment or for real users.
Where you use it
One of the most frequent verbs in the field. You deploy code, a service or a fix, and you deploy to an environment: 'deploy to staging', 'deploy to prod'. Teams also ask 'Who deployed last?' when something breaks.
Examples
I'll deploy the fix to staging first and ask QA to check it.
We don't deploy on Friday afternoons unless it's an urgent security patch.
Usage note
Deploy and release are not always the same: code can be deployed to servers but hidden from users until it is released, for example behind a feature flag.
Common phrases
deploy to production · deploy a fix · deploy a new version · deploy automatically
The act of putting a new version of software onto servers, or one particular time this is done. In Kubernetes it is also the name of an object that manages a set of identical pods.
Where you use it
Common in status updates and tools: 'the deployment failed', 'a deployment is in progress', 'deployment frequency'. Kubernetes users write 'Deployment' with a capital letter when they mean the object.
Examples
The deployment failed because the new container couldn't connect to the database.
Please don't merge anything while the production deployment is running.
Let's scale the Deployment to five replicas before the campaign starts.
Common phrases
production deployment · deployment pipeline · deployment failed · zero-downtime deployment · deployment frequency
An automated series of steps, such as building, testing and deploying, that runs every time code changes and stops if one step fails.
Where you use it
Engineers talk about pipelines all day: they run, pass, fail, turn green or red, and get stuck. A pipeline is made of stages and jobs. Common tools are GitHub Actions, GitLab CI and Jenkins.
Examples
The pipeline failed at the test stage, so nothing was deployed.
Can you re-run the pipeline? I think that error was just a network timeout.
Common phrases
CI/CD pipeline · run the pipeline · the pipeline failed · pipeline stage · deployment pipeline
Short for continuous integration and continuous delivery (or deployment): the practice of testing every code change automatically and delivering it to users through an automated pipeline.
Where you use it
Said letter by letter, 'C-I-C-D'; the slash is not pronounced. It often works like an adjective: CI/CD pipeline, CI/CD tool, CI/CD setup. 'CI' alone usually means the automated build and test part.
Examples
We set up CI/CD with GitHub Actions, so every merge to main goes to staging automatically.
Without a CI/CD pipeline, every release took us a full day of manual work.
Usage note
The 'D' can mean delivery (ready to release, with a manual approval step) or deployment (released to production automatically). Ask which one a team means.
Common phrases
CI/CD pipeline · set up CI/CD · CI/CD tool · CI/CD workflow
A separate copy of the whole system, with its own servers, settings and data, used for one purpose such as development, testing or serving real users.
Where you use it
Typical environments are dev, test, staging and production, often shortened to 'env'. Code moves from one environment to the next. 'Environment variables' are settings passed to an application from outside its code.
Examples
The bug only happens in the test environment, so it's probably a configuration problem.
Each developer can create a temporary environment for their branch.
Usage note
In general English 'environment' often means nature and the planet. In IT it almost always means a system setup, so 'çevre' is the wrong translation here.
Common phrases
test environment · production environment · environment variable · development environment · set up an environment
The live environment that real customers use. A problem in production affects users directly, so changes there are made with the most care.
Where you use it
Usually shortened to 'prod' in speech and chat: 'deploy to prod', 'it's broken in prod', 'prod database'. 'In production' also means 'used for real': 'Kubernetes has been in production here for two years.'
Examples
We found the bug in staging, before it ever reached production.
Never run that script against the production database without a backup.
Common phrases
in production · deploy to production · production environment · production incident · production-ready
An environment set up to be as close as possible to production, used for final testing of a new version before it goes live.
Where you use it
Used as a noun ('it works on staging') or before another noun ('staging environment', 'staging server'). Some companies call it pre-prod or UAT. Data in staging is usually fake or anonymized.
Examples
The release is on staging now. Could you check the payment flow before we go live?
Our staging environment uses smaller servers, so don't trust its performance numbers.
Common phrases
staging environment · deploy to staging · staging server · test on staging
Returning a system to an earlier, working version after a new change causes problems.
Where you use it
The noun is one word, 'rollback'; the verb is two, 'roll back'. Teams say 'do a rollback', 'trigger a rollback' or 'roll back the release'. A good pipeline makes rollback fast and boring.
Examples
Error rates doubled after the release, so we did a rollback within five minutes.
A database migration can make a rollback much harder, so plan it carefully.
Usage note
Write 'rollback' as a noun and 'roll back' as a verb: 'We need a rollback' but 'We need to roll back'.
Common phrases
do a rollback · automatic rollback · rollback plan · trigger a rollback
The process of turning source code into something that can run, such as a binary or container image, usually followed by automated tests. Also the result of that process.
Where you use it
Builds pass or fail, and are 'green' or 'red'. If your change makes it fail, you 'broke the build'. As a verb: 'build the image', 'it builds on my machine'. 'Build time' is how long it takes.
Examples
The build is red again; someone pushed code without running the tests.
Caching dependencies cut our build time from twelve minutes to four.
Common phrases
break the build · build time · the build failed · nightly build · build server
The practice of merging small code changes into the shared main branch often, with each change automatically built and tested so problems are found early.
Where you use it
Shortened to CI. 'CI' also names the system that runs the checks: 'CI is failing', 'wait for CI to pass'. Tools include GitHub Actions, GitLab CI, Jenkins and CircleCI.
Examples
Continuous integration catches most broken merges within ten minutes.
Our continuous integration setup runs unit tests and a security scan on every pull request.
Common phrases
continuous integration server · CI pipeline · CI is green · wait for CI
The practice of keeping code always ready to release, so that any tested version can go to production at the push of a button.
Where you use it
Shortened to CD, together with CI in 'CI/CD'. In continuous delivery a person still approves the final production step; in continuous deployment that step is automatic too.
Examples
With continuous delivery, releasing is a business decision, not a technical project.
We practise continuous delivery, but a product manager approves each production release.
Usage note
Continuous delivery and continuous deployment are both shortened to CD but are not the same; check which one people mean.
A file or package produced by a build, such as a JAR file, a binary or a container image, which is stored and later deployed.
Where you use it
Pipelines produce, upload, store and promote artifacts. A common rule is 'build once, deploy many': the same artifact goes to staging and then to production. Artifacts are kept in tools such as Artifactory or Nexus.
Examples
The pipeline uploads the artifact to the repository, and the deploy job picks it up from there.
We deploy exactly the same artifact that passed the tests on staging.
Usage note
British English writes 'artefact' in general texts, but software tools almost always use the American spelling 'artifact'.
Common phrases
build artifact · artifact repository · upload an artifact · promote an artifact
A release method with two identical production environments: the new version is put on the idle one, then all traffic is switched to it at once.
Where you use it
Blue is usually the current version and green the new one. If something goes wrong, rollback is just switching traffic back. The trade-off is paying for two full environments during the switch.
Examples
With blue-green deployment, our rollback takes seconds instead of half an hour.
The database schema must work with both versions during a blue-green deployment.
Releasing a new version to a small share of users or servers first, watching its metrics, and only then rolling it out to everyone.
Where you use it
Also called a canary deployment; the first small group is 'the canary'. Traffic might go 1 percent, then 10, then 50, then 100. If errors rise, the canary is rolled back before most users are affected.
Examples
The canary release went to 5 percent of users, and error rates stayed flat, so we continued.
Our pipeline stops the canary release automatically if p99 latency goes up by more than 20 percent.
Usage note
The name comes from miners who took canaries into coal mines: if the bird got sick, it warned them of dangerous gas.
A setting that turns a feature on or off without deploying new code, so it can be shown to some users, tested in production or switched off quickly.
Where you use it
Also called a feature toggle. Code is shipped 'behind a feature flag' and the flag is later 'turned on' or 'flipped'. Old flags that nobody removes become technical debt. LaunchDarkly and Unleash are common tools.
Examples
The new checkout is already in production behind a feature flag; we'll turn it on for staff first.
When errors started, we just switched off the feature flag instead of doing a rollback.
Common phrases
behind a feature flag · turn on a feature flag · flip a flag · feature toggle
To release a new version, feature or change to users or servers, often step by step over a period of time rather than all at once.
Where you use it
Very common in release talk: 'roll out to 10 percent of users', 'roll out gradually'. The noun is one word, 'rollout': 'a staged rollout'. In Kubernetes, 'kubectl rollout' commands manage updates.
Examples
We'll roll out the new search to Turkey first and then to the other markets.
The security patch was rolled out to all nodes overnight.
Usage note
Do not mix up 'roll out' (release forward) with 'roll back' (go back to the old version).
Common phrases
roll out gradually · staged rollout · roll out a feature · roll out a patch
Managing servers, networks and other cloud resources by writing configuration files that tools read and apply, instead of clicking through a console by hand.
Where you use it
Shortened to IaC, said 'I-A-C'. The files live in Git, are reviewed in pull requests and applied by a pipeline. Popular tools are Terraform, OpenTofu, AWS CloudFormation, Pulumi and Ansible.
Examples
With infrastructure as code, we can rebuild the whole test environment in twenty minutes.
Nobody changes the firewall by hand anymore; it's all infrastructure as code and goes through review.
Common phrases
infrastructure as code tool · manage infrastructure as code · IaC template · IaC repository
To create and set up computing resources, such as servers, databases or networks, so they are ready to use.
Where you use it
Very common in cloud work: 'provision a database', 'provision infrastructure with Terraform'. The noun is 'provisioning'. 'Overprovisioned' means you have more capacity than you need, which wastes money.
Examples
The pipeline provisions a fresh database for every test run.
It used to take two weeks to provision a server; now it takes two minutes.
Usage note
In general English 'provisions' (plural noun) can mean food supplies or legal clauses; in the cloud the verb means creating resources.
Common phrases
provision infrastructure · provision a server · automated provisioning · overprovisioned
A widely used infrastructure-as-code tool from HashiCorp. You describe resources in configuration files, and it creates, changes or deletes them to match.
Where you use it
Files are written in HCL (HashiCorp Configuration Language). The main commands are 'terraform plan' (show what will change) and 'terraform apply' (make the changes). It keeps track of resources in a state file. OpenTofu is an open-source fork.
Examples
Always read the Terraform plan before you apply it; one wrong line can delete a database.
All our AWS accounts are managed with Terraform, so nothing is created by hand.
Common phrases
Terraform plan · Terraform apply · Terraform module · Terraform state · Terraform provider
Using tools to keep the software, settings and files on many servers in a known, consistent state, instead of changing each server by hand.
Where you use it
Ansible, Puppet, Chef and Salt are the classic tools. Terraform usually creates the servers, and a configuration management tool sets up what runs on them. Ansible files are called playbooks.
Examples
With configuration management, patching 200 servers is one command instead of a week of work.
We use Ansible for configuration management and Terraform for the cloud resources.
A human-readable text format for configuration files that uses indentation and simple 'key: value' lines. Most DevOps tools are configured with it.
Where you use it
Pronounced 'YAM-ul'. Used for Kubernetes manifests, GitHub Actions and GitLab CI pipelines, Ansible playbooks and Docker Compose files. Indentation must use spaces, never tabs, and one wrong space can break a file.
Examples
The pipeline failed because of a single wrong indent in the YAML file.
Run a linter on your YAML before you push it.
Usage note
Files may end in .yaml or .yml; both are correct. The name now officially stands for 'YAML Ain't Markup Language'.
A type of cloud storage that keeps files as objects, each with a unique key and metadata, and lets you read and write them over an HTTP API.
Where you use it
Used for images, videos, backups, logs and data lakes. Amazon S3 is the best-known example; Azure Blob Storage and Google Cloud Storage are similar. Compare it with block storage (disks for VMs) and file storage (shared folders).
Examples
User uploads go straight to object storage, not to the application servers.
Object storage is cheap for large files, but it isn't a replacement for a database.
Common phrases
object storage service · S3-compatible object storage · store files in object storage · block storage
A named container in object storage, such as Amazon S3 or Google Cloud Storage, that holds files and has its own access and storage settings.
Where you use it
You create a bucket, upload objects to it and set a bucket policy. Bucket names in S3 must be unique across all customers. A 'public bucket' that should be private is a common cause of data leaks.
Examples
The logs are archived to an S3 bucket and deleted after ninety days.
Someone made the bucket public by mistake, so we blocked public access and checked the access logs.
Usage note
Azure Blob Storage calls the same idea a 'container', which can be confusing next to Docker containers.
Common phrases
S3 bucket · create a bucket · bucket policy · public bucket · storage bucket
Short for virtual private cloud: your own isolated private network inside a public cloud, where you control IP ranges, subnets, routing and access.
Where you use it
Said letter by letter, 'V-P-C'. AWS and Google Cloud use this name; Azure calls it a virtual network (VNet). Teams create VPCs, split them into subnets and connect them with peering or VPN.
Examples
The database sits in a private part of the VPC and can't be reached from the internet.
We need VPC peering so the analytics account can read from the production database.
Common phrases
create a VPC · VPC peering · inside the VPC · default VPC
A smaller range of IP addresses inside a larger network such as a VPC, used to group resources and control which of them can reach the internet.
Where you use it
Teams talk about public subnets (with a route to the internet, for load balancers) and private subnets (no direct internet access, for databases). Subnets are written in CIDR notation, such as 10.0.1.0/24.
Examples
Put the application servers in the private subnet and only the load balancer in the public one.
We ran out of IP addresses because the subnet was only a /26.
Common phrases
public subnet · private subnet · subnet mask · CIDR range
Short for Domain Name System: the internet's directory that turns names like example.com into the IP addresses computers use to connect.
Where you use it
Engineers create and update DNS records (A, CNAME, TXT, MX). A record's TTL controls how long others cache it, so changes can take time to spread. Engineers joke 'It's always DNS' because so many outages come from it.
Examples
The new site isn't loading for some users because the DNS change hasn't reached their resolver yet.
Lower the DNS TTL a day before the migration so we can switch quickly.
Common phrases
DNS record · DNS lookup · update DNS · DNS propagation · DNS TTL
A copy of data kept separately so it can be restored if the original is lost, damaged, deleted by mistake or encrypted by an attacker.
Where you use it
Backups are taken, scheduled, stored, tested and restored. A backup that has never been restored is not proven to work. The well-known 3-2-1 rule means three copies, on two types of media, with one off-site.
Examples
We take a database backup every night and keep it for thirty days.
Can you back up the config files before you upgrade the server?
Usage note
The noun is one word, 'a backup'; the verb is two words, 'to back up'.
Common phrases
take a backup · restore from backup · nightly backup · backup policy · back up data
A copy of a disk, volume or database exactly as it was at one moment in time, which can be used to restore it or create a new copy.
Where you use it
Snapshots are taken before risky changes and on a schedule. Cloud disk snapshots are usually incremental, storing only blocks that changed. People often say 'take a snapshot before the upgrade'.
Examples
Take a snapshot of the volume before you resize it, just in case.
We restored the database from last night's snapshot and lost only two hours of data.
Usage note
Some snapshots depend on the original disk, so a snapshot is not always a full, independent backup.
Common phrases
take a snapshot · restore from a snapshot · disk snapshot · automated snapshots
Short for identity and access management: the system that controls who or what can sign in to a cloud account and what they are allowed to do there.
Where you use it
Said letter by letter, 'I-A-M'. Its building blocks are users, groups, roles and policies. AWS and Google Cloud call their services IAM; Azure uses Microsoft Entra ID with role-based access control (RBAC).
Examples
The deploy failed because the pipeline's IAM role can't write to the bucket.
We review IAM permissions every quarter and remove anything that isn't used.
Common phrases
IAM role · IAM policy · IAM user · IAM permissions
The security principle that every user, service or program should get only the permissions it really needs for its job, and nothing more.
Where you use it
Usually in the phrase 'the principle of least privilege'. It comes up in security reviews, audits and IAM work: 'this role breaks least privilege', 'scope it down to least privilege'.
Examples
Following least privilege, the reporting service can read the database but never write to it.
Giving everyone admin rights goes against the principle of least privilege.
Common phrases
principle of least privilege · least-privilege access · apply least privilege · least-privilege policy
A sensitive value that a system needs to work, such as a password, API key, token or private key, which must be stored safely and never shown in code or logs.
Where you use it
Secrets are stored in tools such as HashiCorp Vault, AWS Secrets Manager or Azure Key Vault, injected at runtime and rotated regularly. A 'leaked secret' in Git history must be rotated at once, not just deleted.
Examples
Someone committed a secret to the repo, so we rotated the key within the hour.
The app reads its database password from the secret store at startup.
Usage note
A Kubernetes Secret is only base64-encoded by default, not encrypted, so it needs extra protection.
Common phrases
store a secret · rotate a secret · secrets manager · leaked secret · Kubernetes Secret
A virtual firewall in the cloud with rules that say which network traffic, by port and source, may reach or leave a server or other resource.
Where you use it
The name comes from AWS; Azure has network security groups (NSGs) and Google Cloud uses firewall rules. Engineers open or close ports with inbound and outbound rules. 'Open to 0.0.0.0/0' means open to the whole internet.
Examples
The security group only allows traffic on port 443 from the load balancer.
I can't connect to the instance; I think the security group is blocking SSH.
Common phrases
security group rule · inbound rule · open a port · network security group
Turning data into an unreadable form that can only be turned back with the right key, so that people without the key cannot read it.
Where you use it
Cloud teams talk about 'encryption at rest' (stored data, such as disks and buckets) and 'encryption in transit' (data moving over the network, usually with TLS). Keys are often kept in a key management service (KMS).
Examples
Encryption at rest is turned on for every disk and bucket by default.
The auditors asked us to prove encryption in transit between the services.
Common phrases
encryption at rest · encryption in transit · encryption key · end-to-end encryption
The idea that security in the cloud is split: the provider protects the data centers, hardware and core services, and the customer protects their own data, settings and access.
Where you use it
AWS sums it up as security 'of' the cloud (provider) versus security 'in' the cloud (customer). The split depends on the service type: with IaaS the customer manages more than with SaaS. Often mentioned in audits and training.
Examples
Under the shared responsibility model, a public bucket is our mistake, not the provider's.
Every new engineer learns the shared responsibility model in their first security training.
Common phrases
under the shared responsibility model · security of the cloud · security in the cloud
Continuously collecting and checking data about systems, such as CPU use, error rates and response times, so that problems are noticed quickly.
Where you use it
Used as an uncountable noun: 'we need better monitoring', 'monitoring tools', 'add monitoring for the queue'. Common tools are Prometheus, Grafana, Datadog and Amazon CloudWatch. The verb is 'monitor'.
Examples
Our monitoring showed the disk filling up two days before it became a problem.
Before the launch, make sure monitoring is in place for the new service.
Common phrases
monitoring tool · set up monitoring · monitoring system · uptime monitoring · infrastructure monitoring
A record of events written by an application or system, usually one line per event with a time, a level such as INFO or ERROR, and a message.
Where you use it
Usually plural: 'check the logs', 'the logs show…'. As a verb, apps 'log' errors and requests. Teams send logs to central tools such as Elasticsearch, Loki or Splunk and search them during incidents.
Examples
The logs show a timeout every time the service calls the payment API.
Please don't log passwords or card numbers, even at debug level.
An automatic notification sent to engineers when a monitored value crosses a limit or something looks wrong, for example when the error rate is too high.
Where you use it
Alerts fire, trigger, go off and get resolved or acknowledged. Too many unimportant ones cause 'alert fatigue', and engineers learn to ignore them. Alerts often go to Slack, email or paging tools like PagerDuty.
Examples
An alert fired at 2 a.m. because the queue had more than 10,000 messages waiting.
This alert is too noisy. Let's raise the threshold so it only fires for real problems.
Common phrases
an alert fires · alert threshold · acknowledge an alert · alert fatigue · noisy alert
A number measured regularly over time to show how a system is behaving, such as CPU use, memory, request count or error rate.
Where you use it
Metrics are collected, exported, scraped, stored and shown on dashboards. Prometheus and CloudWatch are common tools. Alerts are usually based on metrics crossing a threshold.
Examples
The memory metric has been climbing for a week, which looks like a memory leak.
Let's add a metric for queue length so we can alert on it.
Usage note
Do not confuse 'metric' (a measurement) with 'the metric system' (metres and kilograms).
A screen that shows the most important metrics and charts about a system in one place, so engineers can see its health at a glance.
Where you use it
Engineers build, share and check dashboards, often in Grafana, Datadog or CloudWatch. During an incident people say 'look at the dashboard' or 'the dashboard is all red'.
Examples
I've added the new payment service to the main dashboard.
The dashboard shows a clear drop in traffic at 14:05, right after the deploy.
Common phrases
Grafana dashboard · build a dashboard · monitoring dashboard · on the dashboard
How well you can understand what is happening inside a system from the data it produces, especially logs, metrics and traces, even for problems nobody predicted.
Where you use it
Logs, metrics and traces are often called the 'three pillars of observability'. OpenTelemetry is a common open standard for collecting this data. The short form 'o11y' appears in chats and tool names.
Examples
Monitoring told us the API was slow; good observability showed us exactly which query was the cause.
We're investing in observability before we split the monolith into microservices.
Usage note
Monitoring checks for problems you already know about; observability helps you investigate problems you did not expect.
Common phrases
observability platform · improve observability · three pillars of observability · observability tools
Following a single request as it passes through many services, recording how long each step takes, so you can see where time is lost or errors happen.
Where you use it
Usually 'distributed tracing'. One request produces a trace, made of spans (one per step), linked by a trace ID. Tools include Jaeger, Zipkin and OpenTelemetry. Very useful with microservices.
Examples
Distributed tracing showed that 80 percent of the response time was spent waiting for the stock service.
Add the trace ID to every log line so tracing and logs can be linked.
Common phrases
distributed tracing · trace ID · tracing data · enable tracing
A regular automatic test, often a request to a special URL such as /health, that checks whether a server or service is working and able to take traffic.
Where you use it
Load balancers use health checks to stop sending traffic to broken servers. Kubernetes has liveness probes (restart if stuck) and readiness probes (send traffic only when ready). Health checks pass or fail.
Examples
The health check failed three times, so the load balancer stopped sending traffic to that instance.
Our health check endpoint also tests the database connection, not just the web server.
Common phrases
health check endpoint · failed health check · readiness probe · liveness probe
The ability of a system to handle more users, data or traffic by adding resources, without a big drop in performance or a full redesign.
Where you use it
A key quality in design discussions and job interviews: 'scalability issues', 'built for scalability'. The adjective is 'scalable' and the verb is 'scale'. People ask whether a design 'will scale'.
Examples
Scalability wasn't a concern with 100 users, but at 100,000 the single database became a problem.
We chose a queue-based design mainly for scalability.
Common phrases
improve scalability · scalability issues · horizontal scalability · designed for scalability
A service that receives incoming traffic and spreads it across several servers, sending requests only to the ones that are healthy.
Where you use it
Sits in front of web servers or pods. Engineers put servers 'behind a load balancer' and configure health checks and listeners. Examples are AWS Application Load Balancer, Nginx and HAProxy. Often shortened to LB.
Examples
The load balancer took the broken server out of rotation after three failed health checks.
We put two more instances behind the load balancer before the sale started.
Common phrases
behind a load balancer · application load balancer · load balancer health check · configure a load balancer
Automatically adding servers or containers when demand rises and removing them when it falls, based on rules such as CPU use or number of requests.
Where you use it
Also written 'auto scaling' or 'auto-scaling'. Examples are AWS Auto Scaling groups and the Kubernetes Horizontal Pod Autoscaler. Teams set minimum and maximum limits and a target, such as 60 percent CPU.
Examples
Autoscaling added twelve pods during the match, then removed them when traffic dropped.
Set a maximum for autoscaling, or a traffic spike could double the bill.
The time a request or piece of data takes to travel through a system and get a response, usually measured in milliseconds.
Where you use it
Latency is high or low, spikes and increases. Teams track percentiles such as p95 or p99 latency (the time within which 95 or 99 percent of requests finish), because averages hide slow requests.
Examples
p99 latency went from 200 to 900 milliseconds after the new release.
Users in Asia see higher latency because all our servers are in Europe.
Short for content delivery network: a network of servers around the world that keeps copies of images, videos and pages close to users so they load faster.
Where you use it
Said letter by letter. Cloudflare, Akamai, Fastly and Amazon CloudFront are well-known CDNs. Content is served from 'edge locations'. After a change you may need to 'purge' or 'invalidate' the CDN cache.
Examples
After we put the images behind a CDN, page load time in Turkey dropped by a second.
Users still see the old logo, so please purge the CDN cache.
Common phrases
CDN cache · purge the CDN · behind a CDN · edge location
To handle more load by adding more servers, containers or instances that share the work, rather than making one machine bigger.
Where you use it
Also called horizontal scaling. The opposite, removing servers when load drops, is 'scale in'. Stateless services scale out easily; databases are harder. Autoscaling usually scales out and in automatically.
Examples
During the campaign we scaled out from four to twenty web servers.
The service keeps sessions in memory, so we can't scale out until we move them to Redis.
Common phrases
scale out horizontally · scale out and in · horizontal scaling · scale out the web tier
To handle more load by giving one machine more power, such as more CPU or memory, instead of adding more machines.
Where you use it
Also called vertical scaling. The opposite is 'scale down'. It is simple but has a limit (the biggest machine available) and often needs a restart. Databases are frequently scaled up before they are split.
Examples
We scaled up the database to 32 CPUs, which bought us six more months.
Scaling up is quick, but it doesn't help if the server itself goes down.
Usage note
In general business English 'scale up' just means 'grow'. In infrastructure it specifically means making one machine bigger.
Common phrases
scale up vertically · scale up the database · vertical scaling · scale down
The amount of work a system completes in a given time, for example requests per second, messages per minute or megabytes per second.
Where you use it
Usually discussed together with latency: latency is how long one request takes, throughput is how many requests finish per second. Requests per second is often written RPS. Throughput increases, drops or hits a ceiling.
Examples
After the fix, throughput went up from 2,000 to 5,000 requests per second.
The disk throughput is the limit here, not the CPU.
Usage note
The 'ough' is pronounced like 'oo' in 'too': THROO-put.
Common phrases
high throughput · increase throughput · requests per second · network throughput
The one slowest or most limited part of a system, such as the database or the network, that holds back the performance of everything else.
Where you use it
Teams 'find', 'hit' and 'remove' bottlenecks. Fixing one bottleneck often reveals the next one. Load tests and tracing are common ways to find them. The word is also used for people and processes.
Examples
Adding more web servers didn't help, because the database was the real bottleneck.
The load test showed a bottleneck in the image service at 800 requests per second.
Common phrases
performance bottleneck · find the bottleneck · remove a bottleneck · become a bottleneck
An unplanned event that disrupts a service or makes it worse for users, such as an outage, a big slowdown or a data problem, and that needs a quick response.
Where you use it
Incidents are declared, handled, escalated and resolved. Many companies rate them by severity, such as SEV1 for the most serious. An incident commander coordinates the response, and a postmortem follows.
Examples
We declared an incident when checkout errors jumped to 30 percent.
The incident lasted forty minutes and was caused by an expired certificate.
Usage note
In everyday English an incident can be any event, even a small accident. In operations it means a service problem, not a traffic accident.
Common phrases
declare an incident · incident response · major incident · incident commander · resolve an incident
A period when a service, system or part of it is not working or cannot be reached by users.
Where you use it
Outages can be full or partial, planned or unplanned. You experience, cause, report and recover from an outage. Big provider outages often make the news and appear on public status pages.
Examples
A partial outage in the storage service made image uploads fail for an hour.
The outage started right after the config change, so we rolled it back.
Usage note
In British English 'power cut' is common for electricity; 'power outage' is more American. In IT, 'outage' is standard everywhere.
Common phrases
partial outage · major outage · cause an outage · outage window · unplanned outage
The amount of time a system or service is not available to users, either because of a failure or because of planned maintenance.
Where you use it
Uncountable: 'two hours of downtime', 'minimize downtime'. 'Planned downtime' is announced maintenance; 'zero-downtime deployment' means releasing without interrupting users. The opposite is uptime.
Examples
The database upgrade needs about fifteen minutes of downtime, so we'll do it at night.
Every hour of downtime costs the shop thousands of euros in lost orders.
The time or percentage of time that a system is working and available, often written as a figure such as 99.9 percent over a month or a year.
Where you use it
Engineers talk in 'nines': 'three nines' is 99.9 percent (about 8.8 hours of downtime a year), 'four nines' is 99.99 percent (about 53 minutes a year). Uptime figures appear in SLAs and on status pages.
Examples
We had 99.95 percent uptime last quarter, just above our target.
An uptime monitor checks the homepage every minute from five countries.
Short for service level agreement: a contract that promises customers a certain level of service, such as 99.9 percent uptime, and says what happens if it is not met.
Where you use it
Said letter by letter, 'S-L-A'. Cloud providers often give service credits (a discount on the bill) when they break an SLA. Teams say a provider 'meets' or 'breaches' its SLA.
Examples
The SLA promises 99.9 percent availability, so this outage means we get a service credit.
Our enterprise customers ask for a four-hour response time in the SLA.
Usage note
SLA is a promise to customers with consequences; an SLO is an internal target, usually set stricter than the SLA.
Common phrases
meet the SLA · breach the SLA · SLA credit · uptime SLA
Available outside normal hours to respond quickly if an alert or incident happens. Engineers take turns being on call for their services.
Where you use it
Written with a hyphen before a noun ('on-call engineer', 'on-call rotation') and without one after a verb ('I'm on call this week'). The on-call person 'gets paged' when an alert fires.
Examples
I'm on call this weekend, so I'll keep my laptop with me.
The on-call engineer got paged twice last night for the same disk alert.
Common phrases
on-call engineer · on-call rotation · be on call · on-call shift · on-call handover
A written review after an incident that explains what happened, the impact, the timeline, the root cause and the actions that will stop it happening again.
Where you use it
Good teams write 'blameless postmortems' that focus on systems and processes, not on who made a mistake. Also called a post-incident review or incident report. Ends with action items that have owners.
Examples
The postmortem found that the alert existed but went to a channel nobody watched.
Please write the postmortem by Friday and add an owner to each action item.
Usage note
In medicine a post-mortem is an examination of a dead body (otopsi). In DevOps it only means an incident review.
The deepest, original reason a problem happened, which, if fixed, stops the same problem from coming back, as opposed to its visible symptoms.
Where you use it
Teams 'find', 'identify' or 'get to' the root cause. The process is root cause analysis (RCA), and a common method is the 'five whys'. Many postmortems now list several contributing factors instead of one single root cause.
Examples
Restarting the service fixed the symptom, but we still don't know the root cause.
The root cause was a retry loop that sent ten times more traffic to the database during the outage.
Common phrases
find the root cause · root cause analysis · identify the root cause · contributing factor
Switching automatically or manually to a standby system, such as a database replica or a second region, when the main one fails.
Where you use it
The noun is one word ('failover'); the verb is two ('fail over'). Teams test failover regularly, because an untested one often fails when needed. Returning to the original system afterwards is called failback.
Examples
The database failover took 40 seconds, and most users saw only one error.
We run a failover test every quarter to make sure the second region really works.
Common phrases
automatic failover · failover test · trigger a failover · failover region
Having extra copies of servers, network links, disks or data, so that the system keeps working if one of them fails.
Where you use it
Engineers 'build in' or 'add' redundancy. 'N+1' means one more unit than you need. The adjective is 'redundant': 'redundant power supplies'. The goal is to avoid a single point of failure.
Examples
We added redundancy by running two NAT gateways instead of one.
The network has redundant links, so cutting one cable didn't cause an outage.
Usage note
In British English 'redundancy' also means losing your job because the position is no longer needed. In IT it means extra backup capacity.
Common phrases
build in redundancy · redundant system · single point of failure · N+1 redundancy
Designing a system so that it stays up almost all the time, with no single part whose failure can stop the whole service.
Where you use it
Shortened to HA, said 'H-A': 'HA setup', 'HA pair'. The adjective is 'highly available'. It usually means running in several availability zones with redundancy, load balancing and automatic failover.
Examples
For high availability, the API runs in three zones behind one load balancer.
High availability costs more, so we only use it for customer-facing services.
Common phrases
high availability setup · highly available · HA cluster · design for high availability
The plans, tools and processes for bringing systems and data back after a major event, such as losing a whole data center or region.
Where you use it
Shortened to DR. Two key targets are RTO (recovery time objective: how quickly you must be back) and RPO (recovery point objective: how much recent data you can afford to lose). Teams run DR drills to test the plan.
Examples
Our disaster recovery plan says we must be back online in another region within four hours.
During the disaster recovery drill, we found that two services had no backups at all.
Common phrases
disaster recovery plan · DR site · disaster recovery drill · RTO and RPO
Short for service level objective: an internal reliability target for a service, such as 99.9 percent of requests succeeding over 30 days.
Where you use it
Said letter by letter. An SLO is measured with an SLI (service level indicator), such as success rate or latency. The allowed failure, here 0.1 percent, is the 'error budget'; when it runs out, teams slow down releases.
Examples
Our SLO for the login API is 99.95 percent of requests under 300 milliseconds.
We missed the SLO this month, so the team will focus on reliability work.
Short for site reliability engineering: using software engineering methods to run production systems reliably. It also names the engineer who does this work.
Where you use it
Started at Google in the early 2000s. Key ideas are SLOs, error budgets, automation and reducing 'toil' (repetitive manual work). People say 'an SRE', 'the SRE team' and 'SRE practices'.
Examples
The SRE team owns the on-call process and the incident tooling.
She moved from backend development to SRE because she enjoys debugging production problems.
Usage note
DevOps is a broad culture; SRE is one concrete way to practise it, with clear targets such as SLOs.
Common phrases
SRE team · SRE practices · site reliability engineer · reduce toil
A written step-by-step guide for handling a specific task or alert, such as restarting a service or clearing a full disk, so anyone on call can follow it.
Where you use it
Good alerts link to a runbook. Teams write, update and automate runbooks. 'Playbook' is used in a similar way, often for broader incident procedures. Runbooks are kept in wikis or in the code repository.
Examples
Follow the runbook for this alert: check the queue, then restart the worker pods.
The runbook was out of date, so the on-call engineer lost twenty minutes.
Common phrases
follow the runbook · update the runbook · runbook link · automated runbook
Describes a pricing model in which you pay only for the resources you actually use, for example by the second, hour or gigabyte, with no long-term contract.
Where you use it
The default pricing in the cloud, also called on-demand pricing. Used before nouns ('pay-as-you-go pricing', 'a pay-as-you-go plan'). It is flexible but more expensive per hour than reserved capacity.
Examples
Pay-as-you-go pricing is perfect for our test environments, which only run during office hours.
We started on a pay-as-you-go plan and moved to reserved instances once usage was stable.
Common phrases
pay-as-you-go pricing · pay-as-you-go model · on-demand pricing · pay-as-you-go plan
Reducing cloud spending without hurting performance or reliability, for example by removing unused resources, choosing the right sizes and using discounts.
Where you use it
A regular topic for DevOps teams and finance. Common actions are rightsizing (making instances smaller), deleting idle resources, tagging costs by team and buying reserved capacity. The practice is often called FinOps.
Examples
Our cost optimization work this quarter saved about 20 percent, mostly by deleting idle test servers.
Cost optimization starts with tagging, because you can't fix costs you can't see.
Usage note
British English spells it 'optimisation'; cloud provider documents use the American 'optimization'.
Cloud capacity you commit to use for one or three years in exchange for a large discount compared with normal on-demand prices.
Where you use it
Shortened to RI. The term comes from AWS; Azure has reserved VM instances and Google Cloud has committed use discounts. AWS Savings Plans work in a similar way. Best for steady, predictable workloads.
Examples
Our databases run all year, so we bought reserved instances for them.
Before you buy a reserved instance, make sure the workload won't shrink next year.
Common phrases
buy reserved instances · reserved instance coverage · savings plan · committed use discount
Spare cloud capacity sold at a much lower price, which the provider can take back at short notice when it needs the capacity again.
Where you use it
On AWS, a spot instance gets about a two-minute warning before it is interrupted. Azure and Google Cloud call them Spot VMs. Good for batch jobs, CI runners and stateless workers that can restart safely.
Examples
Our CI runners use spot instances, which cut the bill by about 70 percent.
Don't run the database on a spot instance; it can disappear at any time.
Common phrases
spot instance interruption · spot pricing · run on spot · spot VM
Data leaving a network, for example from a cloud provider to the internet or to another region. Cloud providers usually charge for egress but not for incoming data.
Where you use it
Common in cost and network talk: 'egress fees', 'egress costs', 'egress traffic'. The opposite is ingress (incoming traffic). In Kubernetes, 'Ingress' is also the name of an object that routes incoming HTTP traffic.
Examples
Egress was a third of our cloud bill because users downloaded videos straight from storage.
Putting a CDN in front of the bucket reduced our egress costs a lot.
Common phrases
egress fees · egress traffic · data egress · ingress and egress
Ten questions from this glossary. Pick the right meaning for each term and see your score.
How to learn cloud computing & devops vocabulary
Read real texts from the field every week: cloud provider documentation, public status pages, published postmortems, release notes and the README files of open-source tools such as Kubernetes and Terraform.
Learn each noun together with its verbs: you spin up an instance, roll out a release, roll back a deployment, page the on-call engineer and tear down a test environment.
Practise with your own daily work: write commit messages, pipeline step names, alert descriptions and incident updates in English, even in a Turkish team.
Say abbreviations the way engineers do: SLA, IAM and VPC letter by letter, SaaS like 'sass', YAML like 'yam-ul', and Kubernetes as 'koo-ber-NET-eez' (often written K8s).
Keep a list of pairs that are easy to mix up, such as deploy and release, monitoring and observability, backup and snapshot, SLA and SLO, scale up and scale out.
Frequently asked questions
How can I learn English for cloud computing and DevOps?
Read real texts from the field and notice how words are used together: provider documentation, Kubernetes and Terraform guides, public postmortems and status page updates. Learn each term with its usual verbs and phrases, then use them right away in your own commit messages, tickets, alert texts and incident notes.
Which cloud and DevOps terms should I learn first?
Start with the words of everyday work: deploy, deployment, pipeline, CI/CD, environment, production, staging, container, Kubernetes, instance, region, log, alert, incident and rollback. Then move on to service models, scaling, monitoring and reliability terms, and later to security and cost vocabulary. This glossary follows the same order.
Do AWS, Azure and Google Cloud use the same words?
The core ideas are the same, but product names differ. A virtual server is an EC2 instance on AWS, a virtual machine on Azure and a Compute Engine VM on Google Cloud. A private network is a VPC on AWS and Google Cloud but a VNet (virtual network) on Azure. Learn the general term first, then the name each provider uses.
Should I use the Turkish or the English term at work?
Turkish DevOps teams use many English words as they are, often with Turkish endings: 'deploy etmek', 'pod'u restart etmek', 'pipeline patladı'. Some Turkish forms are just as common, such as 'canlı ortam' for production and 'canlıya almak' for deploying to production. Formal documents, tenders and university courses prefer Turkish terms like 'bulut bilişim' and 'yük dengeleyici', so it pays to know both.
How do I write an incident update in English?
Keep it short and factual: what is affected, since when, what you are doing and when the next update will come. For example: 'We are investigating elevated error rates on the payments API since 14:10 UTC. A fix is being rolled out. Next update in 30 minutes.' Status pages often use the stages Investigating, Identified, Monitoring and Resolved.
