For the whole data science workflow, Vertex AI Workbench serves as a single development environment. The notebook-based environment of Vertex AI Workbench lets you query and examine data, create and train models, and run pipelines of code.
By integrating BigQuery and Cloud Storage, Vertex AI Workbench enables you to access and examine your data from within a Jupyter notebook. Additionally, it uses scheduled executions of your notebook's code that use Vertex AI to automate repeated modifications to your model.
Let us deep dive into Vertex AI Workbench and understand it in depth.
Managed notebooks and User-managed notebooks
The distinctions between managed notebooks and user-managed notebooks are now described, allowing you to select the best option for your project.
For your data science workflow, Vertex AI Workbench offers two Jupyter notebook-based choices.
Managed notebook: Managed notebook instances are Google-managed environments with features and integrations that make it easier for you to set up and operate in a whole production environment based on notebook technology.
User-managed notebooks: Users who require a great deal of control over their environment should use user-managed notebooks, which are Deep Learning VM Images instances that are highly customizable.
Managed notebooks
If you want to utilise a notebook for data exploration, analysis, modelling, or as part of an end-to-end data science workflow, managed notebooks are typically a wise choice. You can carry out workflow-oriented actions inside managed notebook instances without leaving the JupyterLab interface. They also offer a wide range of capabilities and integrations for setting up your data science workflow.
Some of the capabilities and integrations available with managed notebooks are listed below.
From JupyterLab, manage your hardware and framework
In a managed notebooks instance, your JupyterLab interface is where you specify what computing resources, such as how many vCPUs or GPUs and how much RAM, your code will use to run as well as the framework you wish to use. Without leaving JupyterLab or restarting your instance, you can write your code first and then decide how to run it. You can scale your hardware up and down to run your code against additional data quickly for testing purposes.
Custom containers
TensorFlow and PyTorch are just a couple of the popular data science frameworks available in your managed notebooks instance, but you can also install your own Docker container images.
The preinstalled frameworks and your bespoke containers are both accessible from the JupyterLab interface.
Access to data
You don't need to leave the JupyterLab interface to view your data.
You can utilise the Cloud Storage integration to browse data and other files that you have access to from the JupyterLab navigation menu on a managed notebooks instance.
The BigQuery integration allows you to browse the tables you have access to, create queries, see a preview of the results, and import data into your notebook all from within the navigation menu.
Automated notebook runs
A notebook can be programmed to run repeatedly. Vertex AI Workbench will run your notebook file even if your instance is shut down, saving the outcomes for you to view and distribute to others.
Dataproc integration
Running a notebook on a Dataproc cluster allows for rapid data processing. You can execute a notebook file on your cluster after it has been configured without leaving the JupyterLab interface.
User-managed notebooks
Users who demand substantial customization or who need a lot of control over their environment may find user-managed laptops to be a viable option.
Customizable Deep Learning VM instances
Deep Learning VM instances are user-managed notebook instances. When you build a user-managed notebooks instance, you can select particular options for your virtual machine (VM) instance. For instance, when you establish your user-managed notebooks instance, you choose the machine type and the framework. After your instance is created, you can alter the machine type, but doing so necessitates restarting your instance.
You can manually change your user-managed notebooks instance by changing the software and package versions. It takes extra work to modify the framework on your instance.
User-managed notebook instances can be customised in the same ways that Compute Engine instances can be customised because they are presented as Compute Engine instances.
Networking and security
User-managed notebooks may be the best choice for users who have certain networking and security requirements.
In addition to implementing various built-in networking and security capabilities, you may set up a user-managed laptops instance inside of a service boundary using VPC Service Controls. User-managed notebook instances can also be manually configured to meet some unique networking and security requirements.
Notebooks API
You may manage Vertex AI Workbench resources on Google Cloud using the Notebooks API.
google.cloud.location.Locations
google.cloud.notebooks.v1.ManagedNotebookService
google.cloud.notebooks.v1.NotebookService
google.cloud.notebooks.v1beta1.NotebookService
google.iam.v1.IAMPolicy
google.longrunning.Operations
Frequently Asked Questions
What is Vertex AI Workbench?
Users can build and run code that has an effect on their Google Cloud resources using Vertex AI Workbench. The code that users create and run within managed notebooks and user-controlled notebooks instances is their responsibility.
What does the Vertex AI workbench do?
The notebook-based environment of Vertex AI Workbench lets you query and examine data, create and train models, and run pipelines of code.
How many types of notebooks are there in Vertex Workbench?
There are two types of notebooks in Vertex AI Workbench: managed notebooks and user-managed notebooks.
Conclusion
In this article, we have extensively discussed the Vertex AI Workbench.
8+ registered 