<p><strong>NFR computing discussion</strong></p>
<p> </p>
<p>News from Pål</p>
<ul>
<li>Optimistic</li>
<li>2 meetings on mechanism for funding NT1.</li>
<li>Dream was to do it directly through NorCC but NorCC is funded from a treasury line item that can’t be used for infrastructure. Can’t move from one line item to another</li>
<ul>
<li>NorCC is research money and NorLHC is infrastructure money. They can’t be mixed.</li>
<li>Håvard: in the end it doesn’t matter as long as there is funding </li>
</ul>
<li>So still need an infrastructure application to fund NT1. Could be part of the NorLHC3 or a separate application</li>
<ul>
<li>Heidi: some perceive NorCC as a centre and others as project so not yet clear how the NorLHC3 application will proceed</li>
<li>Pål: from the point of view of NFR, NorCC is a project but can be as much as a centre as it wants. But in the end the infrastructure can’t be paid for from NorCC.</li>
</ul>
<li>NeIC isn’t part of the discussion now. Have been in discussion with the other Nordic RCs and are all aware and committed to funding the Nordic T1. At least for the time being want to keep it running as it is. Have also discussed it with NordForsk and they are happy to be the middle man between the funders and the providers. We need to make sure in the 4 countries that we find ways to fund this so that NordForsk don’t have to handle a panoply of applications. Puts a workload on individual research communities, but each community knows its own system. This should be invisible to NordForsk. So far agreements are informal and verbal. In practice NFR will look at the old NeIC MoU and use it as an inspiration for the new one, aim to complete/sign it by summer to have an MoU from 2026-2030</li>
<ul>
<li>Heidi: how does it work: we apply for funds from NFR, it goes to NordForsk and then they buy the equipment?</li>
<li>Pål: need to lock down the details of how this will actually work.  Will money applied for by us go straight to IT-avdelingen? Or will it go via NordForsk. To be understood</li>
<li>Håvard: need to pay for things being done by NeIC, which is a few positions. NordForsk can employ. Need to the money to pay these people</li>
<li>Pål: unclear to what extend this money has been used to cover operational expenses in Oslo and Bergen</li>
<li>Håvard: this hasn’t passed through NeIC. All what is going through NorLHC2 can be unchanged in NorLHC3. The part of NT1 that has previously been covered by NeIC needs to be covered somewhere</li>
<li>Heidi: trying to understand the flow of the funding, administrative overhead and who pays, and positions at UiO/UiB that are paid via NT1 funds</li>
<li>Pål: regarding responsibility for people, they are employed by other employers. NeIC has paid their employers for using those people. That issue is therefore already handled. Administrative overhead: needs to be figured out. Depends on how much it is. Might be a well defined and small additional part to the NorLHC application. RC knows that this part of NorLHC has to be funded.</li>
<li>Heidi: would be useful to see a list of responsibilities between countries and NordForsk so the work is clearly divided, so we can check that it makes sense.</li>
<li>Håvard: hardware funding differs from country to country</li>
</ul>
<li>Need an open discussion about the types of technology and workflows for HL-LHC, for both technical and operational reasons. Could decide to build a common computing centre for all four countries. Need to look at the full cost/benefits</li>
<ul>
<li>James: need to balance abilities of the experiments to use a given hardware type (e.g. GPU) with what is offered</li>
<li>Pål: this needs to be prepared in good time. We can’t continue how we’re doing now but more, this won’t scale</li>
<li>Matthias: transition to a new computing infrastructure is a 5-10 year project, transition period will be very long because the computing infrastructure is already being bought/already exists. It isn’t 1-2 years</li>
</ul>
<li>Summary: NeIC is moving out, we are returning to a model that we had some years ago.</li>
</ul>
<p> </p>
<p>Sigma2 (Matthias)</p>
<ul>
<li>Document: Matthias please provide link. Lists requirements for computing from various wings of the state including research. Particle physics and astronomy is mentioned. </li>
<li>Is this money something extra or is supposed to cover all of our needs</li>
<li>We had this discussion when we applied for NorLHC2 and looked in detail at Sigma2. Need to see if the landscape is different. </li>
<li>Matthias has reached out to Sigma2. Too many loose ends, Sigma2 are in the process of developing and many things are unclear. </li>
<li>Need to evaluate again when we start with NorLHC3. Things are very unclear and no clear entry points. </li>
<ul>
<li>Pål: question is very relevant and no-one can currently answer the question. Wasn’t part of this evaluation process but Sigma2 is quickly moving to a pre-project phase. They have an ambitious plan. We have a chance to follow what is going on and looking at the scope of what they plan to provide. Need to understand if this is just about AI or something else/extra. If we can be concrete about what the future computing needs for HL-LHC are this can be useful. </li>
<li>This will be part of the MoU</li>
</ul>
<li>James: reminder that we have to provide raw archival of data to be designated as a T1</li>
<li>Heidi: reminde that T1s have to guarantee 24-7 operation. </li>
<li>Matthias: Sigma2 can’t yet guarantee 24-7. Quite strange given the things they are supposed to be supporting</li>
<li>Matthias: seems to be some momentum now. In particular we need to get together to give Pål the information needed to compose the MoU.</li>
<li>Pål: the MoU will not be a 20 page document, it will be much shorter. We should work together and discuss from the bottom up about what we need/want</li>
<ul>
<li>Heidi: this needs to be intensified in the computing</li>
</ul>
<li>Matthias: Sigma2 might consider to provide OpenStack infrastructure, something like NREC or even integrated to NREC, that would be similar to what we</li>
<li>are running today. </li>
<ul>
<li>Should evaluate use of the existing HPC infrastructure </li>
</ul>
<li>Heidi: should we prepare to add T1 funding to NorLHC3 application?</li>
<ul>
<li>Pål: yes! Issue can be treated separately so allow continuation of funding. Not necessary to mention Sigma2 at this stage</li>
<li>Matthias: how much should we press Sigma2 on price/feasibility study?</li>
<li>Pål: this can be part of the preparation for the MoU</li>
</ul>
<li>Matthias: NorLHC2 is until the end of 2027, but for NT1 we need funding from 2026. </li>
<ul>
<li>Pål: This is why we start the NorLHC3 application now and be fast-tracked. NordForsk don’t need the cash Jan 1st. </li>
<li><strong>NorLHC3 application needs to be done this year </strong></li>
<li>Next one will be late in 2027 which is too late</li>
</ul>
<li>Heidi: people are activated to start the application. Starting point is to assume we continue as before, separately from SIgma2, but put in the funding a path to make a transition. Should maybe make a quick check with Sigma2, but since they are under development themselves they may not be able to give so much information.</li>
<li>Pål: should set up a meeting with Sigma2, Pål would be happy to join. Needs our help to understand needs and alternatives. </li>
<li>Heidi/Pål: follow up end of February</li>
</ul>