Cloud & Compute Glossary

Bring Your Own Cloud allows users or organizations to connect and run software or tools within their own existing cloud environments (e.g., their AWS or GCP accounts), rather than using the vendor’s infrastructure.

Carbon-aware computing means making decisions (e.g., when or where to run workloads) based on real-time data about the carbon intensity of electricity. This helps reduce emissions by shifting workloads to greener energy sources or regions.

A hybrid cloud combines on-prem, private, and public cloud environments. Workloads can shift between environments for flexibility, cost efficiency, and regulatory compliance, enabling organizations to use the best of each.

An instance typically refers to a single virtual machine (VM) or containerized environment running in the cloud or on-premises. It provides compute resources (CPU, memory, etc.) and can host services, applications, or runners. The virtual machine or container the runner runs on.

A compute job represents a discrete task that a computer system needs to complete. It may involve processing data, executing code, running algorithms, or performing automated tasks.

In the context of GitHub Actions, a compute job is a specific job defined within a workflow, executed on a runner, and responsible for carrying out a part of the automation process — such as installing dependencies, running tests, or deploying code.

The lowest-carbon region refers to the geographic cloud region currently generating electricity that is most renewable and least carbon-intensive — meaning the lowest CO₂ emissions per kilowatt-hour (kWh) at that moment.

For example, us-west1 (Oregon) is often one of Google Cloud’s lowest-carbon regions due to its strong reliance on hydro and wind power.

Multi-cloud refers to the strategy of using services from multiple cloud providers—such as AWS, Google Cloud Platform (GCP), and Microsoft Azure—within a single architecture. It allows organizations to avoid vendor lock-in, optimize for cost, performance, or sustainability, and improve reliability by distributing workloads across different cloud environments.

Multi-region refers to deploying and running applications across multiple geographic regions or data centers. This enhances availability, performance, and resilience—and enables carbon-aware choices based on regional energy emissions.

On-prem refers to computing infrastructure that is physically located within an organization’s facilities, as opposed to being hosted in a public cloud. The organization owns and manages the hardware and data centers.

Orchestration automates where and how jobs run across infrastructure — balancing performance, cost, and emissions. Kubernetes orchestrates containers across clusters; CarbonRunner orchestrates workloads like CI/CD and AI across cloud regions.

A private cloud is a cloud computing environment exclusively used by one organization. It can be hosted on-prem or in a third-party data center, offering greater control and privacy but requiring more maintenance than public clouds.

A public cloud is a cloud infrastructure offered by providers like AWS, Google Cloud, or Azure. It’s shared among multiple customers, who rent compute/storage/networking resources on-demand. It offers scalability, ease of use, and reduced overhead.

A runner is an agent or process that executes jobs or workflows, such as builds, tests, or deployments, in CI/CD pipelines. It pulls instructions from a CI/CD system (like GitHub Actions or GitLab CI) and runs them on a specific machine or container.

A virtual machine (VM) is a software-based simulation of a physical computer. It runs its own operating system and applications, just like a real machine, but is hosted on shared physical hardware using a hypervisor.

VMs allow multiple isolated computing environments to run on a single physical server — making them useful for scaling workloads, testing, and securely running different systems in parallel.

A workflow is a GitHub Actions automation file that defines a set of jobs arranged in sequence or in parallel, outlining the full automation process for software delivery (e.g., build → test → deploy). Workflows are triggered by events like code pushes, pull requests, or scheduled times, and are stored in the .github/workflows/ directory.

A workload is any computing task that needs to be processed - whether it's running an application, handling data, or performing calculations. Every workload consumes specific computing resources including CPU, memory, storage, and network capacity. In cloud computing, a workload refers to any service, function, or application running on cloud infrastructure to perform a specific task or deliver a capability.

In the context of GitHub Actions (or other automation platforms), workloads can include individual jobs within a workflow. These workloads might take the form of a CI build step, a containerized microservice, a data processing script, a scheduled backup task, or even a machine learning training run — essentially, any task that consumes compute resources to perform a defined function.

CarbonRunner is an intelligent carbon-aware workload orchestrator that routes jobs—like CI/CD builds, AI model training, Data pipelines, Batch processing and render jobs — to the lowest-carbon region in real time.

It supports multi-cloud, hybrid-cloud, and BYOC environments.