How to manage many users using distinct notebooks ?

[Cookistador]

HI Everyone,

I'm helping a friend organize a small class(at school) with Fabric, but I have a question about the setup for notebooks.

The idea is to use a small license (like an F4), but I'm concerned about how notebooks will perform. To minimize costs, the setup I have in mind is:

The minimal number of nodes for a notebook is 2 (I’ve encountered issues when this value is set to 1).

Set the node size to "small" (4 cores).

This means that just to execute one notebook, it will require 8 Compute Units (CUs). With an F4 license, they wouldn’t be able to run more than one notebook per user. Is that correct?

The goal is to create small notebooks to demonstrate to students how to perform transformations. Do you see any alternatives? I considered High Concurrency Mode, but it seems to work only for a single user.

Additionally, I’ve noted the session-sharing conditions:

Sessions must be within a single-user boundary.

Sessions should share the same default lakehouse configuration.

Sessions must have identical Spark compute properties.

How do companies with 10-15 people working on notebooks typically manage this? Do they generally stick with Databricks instead of Fabric?

Many thanks for your guidance! 😊

[Poojara_D12 · Accepted solution]

Hi @Cookistador 

For a classroom setup where multiple students need to run notebooks simultaneously in Microsoft Fabric, the configuration you’re considering (using an F4 license with minimal compute resources) could become quite restrictive. Since each notebook execution requires 8 Compute Units (CUs) with the setup you mentioned (2 nodes with small size, 4 cores), and the F4 license limits the number of users per session, this would indeed mean that only one notebook could be executed per user at a time, making it challenging for multiple students to run their own notebooks concurrently. To minimize costs while still supporting small notebooks for students, there are a few alternatives you can consider. One option is to explore High Concurrency Mode, which allows shared resources but is typically more suited for single-user environments. Another approach is to reduce the compute resources by using smaller nodes with fewer cores to lower the CU requirement, though this might compromise performance. If you need multiple users to work at once, manual scaling could help, though it would require monitoring and adjustments based on usage. In some cases, Databricks might be a more suitable choice for classrooms with multiple students, as it’s designed to efficiently handle multiple users simultaneously with collaborative notebooks, shared Spark clusters, and better management of concurrent workloads. Databricks is a popular solution in companies, especially those with larger teams, as it supports shared compute clusters that scale well across teams. It could provide a more cost-effective and flexible option for your scenario compared to Fabric. Additionally, if you choose to stick with Fabric, setting up separate workspaces for each student could help manage resource allocation more effectively, although it may introduce some complexity.

[v-kpoloju-msft]

Hi @Cookistador,

Thank you for reaching out to the Microsoft Fabric Community Forum.

We really apologies for the inconvenience, after reviewing the issue of manage many users using distinct notebooks, please go through the following steps.

• Setting the minimal number of nodes for a notebook to 2, using small nodes 4 cores, is a suitable starting point. This configuration ensures you have the necessary compute resources to run the notebooks efficiently.
• High concurrency mode offers advantages but also has its limitations. It is intended for multiple users running jobs simultaneously however, you observed that it appears to function only for a single user. This may be problematic if each student requires an isolated environment.
• The conditions you specified single user boundary, same lake house configuration, identical spark compute properties are essential for session sharing. Adhering to these criteria can significantly enhance resource management efficiency.
• As you correctly noted, an F4 license may pose difficulties for running multiple notebooks concurrently. If the objective is for each student to operate their own notebook, it may be necessary to consider upgrading to a higher capacity license.
• Many organizations utilize platforms like data bricks for managing multiple users working on notebooks. Databricks provides features specifically tailored for collaborative data science and engineering tasks, which may better suit your requirements.

Also, please go through the below following link for better understanding:
Microsoft Fabric concepts - Microsoft Fabric | Microsoft Learn

If this post helps, then please give us Kudos and consider Accept it as a solution to help the other members find it more quickly.

Thank you.

[v-kpoloju-msft]

Hi @Cookistador,

May I ask if you have resolved this issue? If so, please mark the helpful reply and accept it as the solution. This will be helpful for other community members who have similar problems to solve it faster.

Thank you.

[v-kpoloju-msft]

Hi @Cookistador,

I wanted to check if you had the opportunity to review the information provided. Please feel free to contact us if you have any further questions. If my response has addressed your query, please accept it as a solution and give a 'Kudos' so other members can easily find it.

Thank you.

[v-kpoloju-msft]

Hi @Cookistador,

I hope this information is helpful. Please let me know if you have any further questions or if you'd like to discuss this further. If this answers your question, please Accept it as a solution and give it a 'Kudos' so others can find it easily.

Thank you.

[GilbertQ]

Hi @Cookistador 

I would recommend leaving the default spark configurations because depending on how much your notebook is doing will relate to how many capacity units will consume with. regards to the other sharing conditions, you should be fine to allow multiple people to edit the same notebook. I would rather start with a few people just to see how it initially goes, just to make sure, but in my experience it should work.