Glossary of Terms

  • API: “Application Programming Interface”. Interface for developers or coders to interact with a system, library, or framework. It provides developers with standard commands for performing common operations so they do not have to write the code from scratch.

  • Cluster: A group of computers that work together that perform a similar function or collaborate on the same function.

  • CPU: “Central Processing Unit”. A component within a computer that receives instructions and follows them to process data or information. Contains at least one core (ALU), control unit, and memory cache.

  • Credential: One of a variety of digital passcodes or methods of authentication to confirm the identity of a user or computer. Should be stored in secure formats, not plain text.

  • CUDA: “Compute Unified Device Architecture”, though the acronym is the only term used today. A parallel computing platform and application programming interface (API) model created by NVIDIA. The GPUs usually used in Saturn Cloud applications have NVIDIA processors with the CUDA framework.

  • Dashboard: A user interface displaying visualizations and/or data with realtime or near realtime updating. May be created and “served” by a number of different tools, including Python libraries Voila and Panel. More information.

  • Dask: A parallel computation framework for Python. https://dask.org/

  • Dask Cluster: A group of computers that work together to run workloads using Dask. Enables parallelization across multiple machines.

  • Dask Worker: One computer in a Dask Cluster, specifically assigned to complete calculations.

  • Dask Scheduler: One computer in a Dask Cluster, specifically assigned to pass instructions to Dask Workers. Usually only one per cluster is needed.

  • Delayed Function: A function designed to be defined at one point, but not actually run until a later time. See also “Lazy Evaluation”.

  • Deployment: Software process that makes a tool or service available to users. In Saturn Cloud, refers to the different types of tool or service that can be made available to users through the application, such as a Dashboard.

  • Eager Evaluation: Opposite of Lazy Evaluation. The name for a strategy where functions are defined, and run instantly at the time of defining.

  • EC2: EC2 is the shorthand for Amazon Web Services (AWS) Elastic Cloud Computing service. Through this service, users can rent computation resources for a period of time, and pay for use for only as long as they need it. Our customers use these instances to get compute resources on Saturn Cloud. https://aws.amazon.com/ec2/

  • Git: Software for tracking changes in files, usually code. Allows users to easily share code and see the differences between versions of one file. Commercial tools where users can access git include Github, Gitlab, and Bitbucket.

  • GPU: “Graphics Processing Unit”. Alternative to CPU. A computer processor optimized for rendering graphics, but also very effective for highly parallel machine learning code.

  • Job: An aggregation of tasks or functions that one intends to run on a computer or cluster.

  • Jupyter Lab: A full featured user interface for interactive computation in Julia, Python, or R. It offers multiple tools besides the Jupyter Notebook, including terminal, text editor, file browser, visual widgets, rich outputs, etc.

  • Jupyter Notebook: An interactive computational environment, designed specifically for creating and running interactive code in Julia, Python, or R languages.

  • Jupyter Server: A server that hosts a Jupyter process, which serves up 2 applications, Jupyter Lab and Jupyter Notebook inside the Saturn Cloud product.

  • Lazy Evaluation: The name for a strategy where functions are defined at one point, but not actually run until a later time. See also “Delayed Function”.

  • Local Cluster: A collection of processes on a single computer that run workloads using Dask. Enables parallelization on one machine.

  • Notebook: An interactive computational environment. For Saturn Cloud, this usually refers to Jupyter Notebook, which is designed specifically for creating and running interactive code in Julia, Python, or R languages.

  • Parallelization, Parallel Computing: Computation where many calculations or the execution of processes are carried out simultaneously. May be conducted on one computer or clusters of many computers. In Saturn Cloud, Dask is the framework we use to enable parallelization.

  • Prefect: An open source workflow orchestration tool made for data-intensive workloads. This allows you to schedule and organize jobs to run any time, in your chosen order. It can accommodate dependencies between jobs, and is very useful for data pipelines and machine learning workflows. More information.

  • Project: In Saturn Cloud, a project is a collection of a few tools that work together. A project always contains, at minimum, one Jupyter Server. It may also contain a Dask Cluster, one or more Deployments, or other things. More information.

  • Scheduler: One computer in a computing cluster, specifically assigned to pass instructions to Workers. Usually only one per cluster is needed.

  • Spot Instance: A Spot Instance is a type of AWS EC2 machine that you can request. It is heavily discounted off the standard price, but may not always be available as it depends on where excess capacity exists. Learn about EC2 Spot Instances on AWS website.

  • vCPU: Virtual CPU. The equivalent computing power/resources of a CPU, but distributed across multiple actual pieces of hardware. Allows developers to use the computing resources in time slots, sharing with other users. A vCPU represents the computing power of a CPU, across multiple resources.

  • Worker: One computer in a computing cluster, specifically assigned to complete calculations.




Need help, or have more questions? Contact us at: We'll be happy to help you and answer your questions!