1. InfinStor Transforms

1.1. Concept

A Transform represents a packaging of

• python code
• and its environment ( conda or docker environment ), so that the packaged code can run in different execution environments, including many public clouds (like AWS, Azure, GCP), JupyterLab and others.

1.2. Developing Transforms

InfinStor includes support for developing transforms from provided templates or by modifying existing Transforms

1.2.1. Transforms Types

InfinStor transforms are of two types:

• Transforms with no InfinStor managed Data input - More information here Transforms with no Input for more details
• Transforms with InfinStor managed Data input - More information here Transforms with InfinStor Managed I/O for more details

1.3. Running Transforms

Transforms can be executed in the following locations:

• Inline in the jupyterlab environment
• In a Single VM in the cloud (AWS, GCP, Azure, others)

1.4. Develop Transform Graphs

A Transform Graph

• represents a pipeline or graph of tasks that need to be executed from start to finish
• with any number of intermediate tasks, and with any number of connections (flow of data from one task to another) between the tasks

For Machine Learning and Deep Learning workloads, the Transform Graph can be used for

• preprocessing data,
• training models,
• inferencing using trained models
• and other complex processing needed in machine learning pipelines

For details see Create a Transform Graph

1.5. Running Transform Graphs

After a transform graph is created, as described above, Run Transform Graph describes how to run the created transform graph

1.6. User Interface

The primary GUI to access the above transforms functionality is the InfinStor Jupyterlab Sidebar