RSP Future Functionality FAQ

This page contains answers to frequently asked questions (FAQ) about the future full functionality of the Rubin Science Platform (RSP).

Will batch resources be available?

Yes.

As described in Section 4.3 of LSE-319, the future RSP will provide a shared computing cluster for users to perform analysis through its Portal, Notebook, and Web API Aspects. The cluster will be managed by a workload management system based on predetermined policies to allocate resources to individual users or groups. RSP users will have an initial allocation of batch CPU time, with the possibility of applying for additional quota through a Resource Allocation Committee. The size of the batch computing resource is determined by the requirement to provide 10% of the total LSST computing and storage resources to users. Users will be able to launch jobs primarily through the APIs exposed in Notebook and Web API Aspects, and the Portal may potentially use the batch computing cluster too.

References:

  • LSE-319: Science Platform Vision Document

See also:

  • DMTN-202: Use cases and science requirements on a user batch facility

  • DMTN-223: User batch - possibilities and plans

Will users be able to install custom software?

Yes.

In fact, in the current DP0 era of the RSP, users are already able to install custom software into their own homespace. In addition, during the operations-era RSP there will be a process for users to request certain popular packages to be added to the common environment for use by all.

References:

Will GPUs be available in addition to CPUs?

No plan at this time.

As of 2023, there are no plans to include GPUs as part of the resources made available on-project (including “bring-your-own” resources). At this time, although there are early discussions about providing a mechanism for users to “bring their own” Google resources to the RSP, including GPU virtual machines, users interested in GPUs should contact the Independect Data Access Centers program (see, e.g., Contributed Computing Resources).

Definitions:

  • GPU: graphics processing unit

  • CPU: central processing unit

Will users be able to download data?

Yes, for limited subsets of the data.

However, it will be impossible (and impractical) to download the full amount of data that the LSST will produce. The RSP, in particular its Notebook Aspect, are designed to provide the users with resources at the LSST Data Access Centers such that computation and analyses can be performed “next to the data”.

Will users be able to work offline?

No.

The full functionality of the RSP Aspects will require an online connection in order to access the LSST data and perform analyses on them.

Will use of IDEs be enabled?

IDEs such as VSCode support connections to a JupyterLab server. However, this currently doesn’t work with the Notebook Aspect of the Rubin Science Platform. Unfortunately, it’s challenging to make this work in our environment at the moment, and this feature is not currenyly a priority in the current RSP developement.

Definitions:

  • IDE: integrated development environment (e.g., VSCode, DataSpell)

Will users be able to share within groups?

Yes.

The future RSP will support collaborative work, and users will be able to link accounts into groups in order to share workspaces (Section 3.6, LSE-319). Users and groups “will be able to store selected or transformed subsets of the LSST dataset, or upload related datasets for joining to the LSST dataset” (Section 4.1, LSE-319).

Additonal Questions

To ask additional questions, please use the “Support – Rubin Science Platform” category in the Rubin Observatory Community Forum