Allowing Multiple Environment Creation in the Same Account-Region about dataall HOT 2 OPEN

Please @noah-paige @petrkalos @SofiaSazonova @anmolsgandhi have a look. To provide some more context: the main pain point that we are trying to solve is that for users it is difficult to navigate data.all UI based on the use-cases they work on. At the moment an environment is only an infra-link to the AWS account, but what @sandeephs1 and team want to achieve is some sort of logical grouping of resources so that when a user logs in, they can directly see the use-cases they have access to and the datasets associated to that use-case (in the diagram Price, inventory are use-cases). At the moment they have designed this logical grouping based on Environments, but it is not possible to have more than 1 environment per AWS account region, because of the infra that is deployed (mainly the auto created pivot role).

Option 1: Hack the Environments

Separate the pivot role stack somehow from the environment stack and create dependencies between the environments in the same account-region so that when one is deleted the pivot role stack remains to serve the others.

• (+) No new concepts
• (-) dependent infra = not very robust
• (-) duplicated infra = service quotas ... (e.g. SageMaker domains)

Option 2: Introduce the concept of data.all Use-Case/Project

Create a new data.all construct that groups resources together. It does not contain any infra, it is just a logical grouping of data.all resources. In this case I think we can even go a step further and not limit it to datasets, but add also notebooks....

• (+) attacks the pain point root cause
• (+) extensible
• (-) new concept
• (-) requires design, it is not a quick fix

from dataall.

@dlpzx this one is not clear to me. Could we make it more detailed an explain what the desired solution is, how it would look in data.all. It sounds like what is being asked is just a new entity to group environments together.. But how is that grouping actually going to be used in practice? If that's all that's needed you can just tag them with the same TAG and find them that way.

I'd like more clarity on what is being asked for and why..

from dataall.