5: Outstanding Infrastructure by

General comments

2023-12-04 16:22:37 UTC

Do we have the resource to make these happen?

Some resource investments are starting to make progress already.

Federation across UKRI funders

2023-12-04 16:23:33 UTC

Introduce ability for IRIS to accept funds from other research councils where appropriate (e.g. management of H&ES funding)

Federation - lacking from 5A

ucapjay — 2023-12-04 16:24:09 UTC

IAM was the first step in Federation. However over the last 2 years, it has become clear that File Transfer, data management, remote job execution are now needed

Sustainable, predictable funding

2023-12-04 16:24:10 UTC

Why is IAM singled out?

2023-12-04 16:24:12 UTC

Don't we need IRIS resource integrated with a full suite of federation enabling services - accounting, Helpdesk, etc. etc.

Dedicated funding for retiring resources sustainably

2023-12-04 16:24:31 UTC

Support and development of the career pathway and a sustainable pipeline of skilled people to support outstanding infrastructure (e.g. training, visability)

2023-12-04 16:25:49 UTC

How to make prototypes more widely available to community

2023-12-04 16:26:15 UTC

Links to federation activities to make them more visible.

2023-12-04 16:26:41 UTC

I would probably put Security top.

We have done a lot over the last 20 years for GridPP/EGI concerning distributed infrastructures - and we should ensure IRIS learns from this. Not just security but with other things.

Prioritising the priorities

2023-12-04 16:27:25 UTC

5B, sustainability, is a big change in culture so will need a lot of effort to make the changes

5A lives in parallel with 5C; in both cases we have to put these at the foundation of our work, especially as we grow the scope of IRIS to include other communities

(this group was very security heavy, so acknowledging that bias)

Support for effort to deliver prototypes and publicise them

2023-12-04 16:27:33 UTC

Need a PRISM to manage this - PRISM is a job category at UCL (see PRISM comment in challenges column) which is ideally suited to this kind of role.

Also relates to ExCALIBUR Hardware and Enabling software programme.

Ongoing funding line for prototypes at major sites

2023-12-04 16:27:48 UTC

Enabling new communities to explore new hardware / innovative testbeds

2023-12-04 16:30:32 UTC

People with the domain knowledge and communities to interface with the technical knowledge to understand if it is appropreate to test an appropreate testbed.

This might be

eg. https://www.pris-managers.ac.uk/

Community needs to have strategic sight of future directions

2023-12-04 16:35:23 UTC

Both to feed into their definition and to prepare for what is coming.

Accessibility

2023-12-04 16:35:41 UTC

Extra information that would be available to existing communities; eg for GitHub you have metrics about how often updates are made, how are issues resolved. What other metrics would be useful for IRIS?

Need to include international angle in terms of strategic planning

2023-12-04 16:35:53 UTC

The way the IAM priority is phrased makes it sound like a hammer to which everything is a nail.

2023-12-04 16:35:55 UTC

2023-12-04 16:38:35 UTC

Need to catalogue assets to make sure that the availability of resources in IRIS is clear (which also helps to secure things).

Tangentially, how fine-grained can accounting be usefully made; job meta-data - how detailed do we need our taxonomies?

How about: 5E, 5B, 5A, 5C, 5D?

2023-12-04 16:38:38 UTC

Priorities:

2023-12-04 16:38:57 UTC

5E: flexibility underpins how IRIS infrastructure should operate - IRIS should provide infrastructure components for community to pick and choose, rather than just be a singular blob.

5A: this is ONLY second if it is a full federating objective, and not just IAM. IRIS needs full federation adoption - this must include other federation services, such as accounting services like GocDB and APEL.

5C: Not much needs to be said about prioritising security

5D: funding to drive innovation can drive huge rewards within infrastructure development

5B: This is still very important, but we had to put one last 🤷‍♂️

Rent out experimental or excess hardware to other uses (e.g. H&ES)

2023-12-04 16:39:23 UTC

Contributing to and learn from best practice in running supercomputing centres internationally?

2023-12-04 16:39:44 UTC

A global effort to bring together public computing centers
for the advancement of science and research

Supercomputing Centers

Resourcing

2023-12-04 16:40:12 UTC

For all of this the challenge is sufficient funding to run the core functions that make the various deployments function as a joined up infrastructure - as well as the operational funding for resources deployed at member institutes.

5A is too focused on IAM and not on Federating as a whole.

2023-12-04 16:41:44 UTC

The AAI, implemented through IAM, is a key underpinning component that drives federation but through just focussing IAM we risk not developing a full federated infrastructure. The accounting and policy components are still required.

H&ES at well defined TRLs

2023-12-04 16:42:48 UTC

e.g. stuff that is near production ready, and stuff that is crazy. Perhaps low TRL stuff should be funded via research grants rather than H&ES

IRIS RSAP form tailored for development/ innovation/ testbeds

2023-12-04 16:55:38 UTC

The computational model and usage predictions may be unknown and harder to quantify when science communities are testing/evaulating innovative testbeds and hardware.

Additional guidance on applying to the IRIS RSAP template and how to deal with the unknowns.

Bring H&ES into IRIS

2023-12-04 17:00:49 UTC