As mentioned in several articles, such as Creating a Notebook and Launching a Notebook, Syntasa notebooks have a kernel running that can run a variety of functions within the notebook, but at times you may need a bigger cluster or custom libraries loaded that will require a Syntasa runtime to be attached to the notebook.
The notebook functionality that is available without a runtime attached depends on the version of your Syntasa environment. As of Syntasa 7.0, the standalone notebook functionality has been expanded and thus the circumstances requiring an attached runtime have reduced.
Pre-Syntasa 7.0
Prior to Syntasa 7.0, the available functionality in a notebook without an attached runtime is limited to standalone Python code. If you need to execute Spark code within your notebook then a runtime needs to be attached and running.
Post-Syntasa 7.0
In Syntasa 7.0 the notebook functionality has been expanded so users can perform most experiments without needing to attach and start a runtime to the notebook. This has been accomplished by automatically attaching a small Kubernetes cluster to each notebook.
You can now execute standalone Python code and Spark code in your notebook without needing to attach a runtime.
When running large experiments or you find the default notebook's performance is degrading, then a runtime can be attached. Once a runtime is attached and running, all notebook execution will utilize this runtime and no longer use the default cluster.