Compute Settings

Stateful Compute

For code execution zerve uses cloud compute (lambda, fargate, GPU or Kubernetes) at block level. When a block is run, zerve spins up relevant compute resources for code execution.

Once successfully executed the results are cached, serialized and stored on disk. This is built on top of a canvas environment for true collaboration and uses DAG workflow to process code execution order.

Post successful code execution data is stored on disk separate from compute. This creates a stable environment to do data science and removes all the shortcomings that we get from a normal jupyter notebook.