🐾What's new in updated SageMaker Studio🐾
🤓 Have you already tried the new SageMaker Studio experience? Wonder how does it differ from the previous version? Should you upgrade to the new version and what can you get from it?
Main changes
New SageMaker Studio is not just an updated UI for the previously well-known SageMaker Studio. AWS presented a different vision for the development process instead — SageMaker Studio has become a control plane for your ML development. Also, now you have more options for IDE, but you still can use the familiar version of SageMaker Studio which was renamed to Classic.
New features
Several IDE options
SageMaker Studio Classic — the previous version of Studio with all well-known features like Pipelines, Model Registry, etc.
JupyterLab for the data scientists who love classic Jypyter notebooks
Code Editor for ML engineers who prefer to work with Visual Studio Code
RStudio — development environment for R
SageMaker Canvas for zero-code model development
Shared spaces for collaborative work
With shared spaces, you can collaborate with your teammates in notebooks in real time. Previously, you didn’t have such an opportunity and shared the results of your work via the S3 share option.
Unified models management
Now you can track model packages in SageMaker Studio UI as deployable models in the Models section, which previously were available only using the AWS console.
Unified running resources control
With the new Studio experience, you can control running instances for all applications in a single tab. It is possible to see which application uses instance and details regarding instance type, and storage size.
Pricing
There are no changes to the pricing of the SageMaker Studio experience renamed to Classic - you still pay for resources used to run notebooks, pipeline steps, and endpoints. For the new applications such as Code Editor, pricing is quite similar and you pay only for the resources you use.
My impression
When considering ML development in the AWS cloud, I often compare SageMaker with Databricks. Databricks previously held an advantage with its real-time notebook functionality, crucial for data scientists and ML engineers. Now, SageMaker has even more usability with shared and private spaces.
Constantly switching between Studio UI and AWS UI to find ML resources was frustrating sometimes, so I like how all the information is ogranised in a new Studio. I tried Code Editor IDE and I would say it’s pretty convenient, but I personally still love coding in PyCharm😁
Thank you for reading, let’s chat 💬
💬 Do you like the new applications in SageMaker?
💬 What is your general impression about the introduced changes?
💬 Have you found any bugs in applications?
I love hearing from readers 🫶🏻 Please feel free to drop comments, questions, and opinions below👇🏻
