 Good afternoon. Welcome. And thank you for attending this session. I am Dale Robertson. I work for JISC in the UK. And I'm also the task leader of Task 2.3 OVOSCUB, which is the Governance and Sustainability Task. Next slide, please. So just to have a quick look at the objectives for this session. I'll just clear my screen so I can actually see. There we are. So the idea is to share some experiences in provision and consumption of services across borders and discuss some of the issues that have already been encountered by people with experience of this. But also to collect some audience input from you as a contribution towards the ongoing work that we have in this area within the OVOSCUB project. Next slide, please. So we've got what looks like a very busy agenda, and I hope it will be very interesting for you. We've got a brief introduction from me to give you some scene setting followed by three 10-minute presentations and then a panel session during which, of course, you have the opportunity to offer your questions and comments. Next slide, please. Some brief housekeeping. The main thing that I wanted to say was that what I'd like to try to do by way of organization for this session is to use the Zoom chat only for technical issues and to use Slido for Q&A, so for your questions and for your comments relating to the actual subject matter of the session. So if you can go to Slido, which I hope is becoming pretty familiar to you by now, put in the code, the EOSCUB week code, and then select in the drop-down this session, so 19th of May issues in cross-border, etc. Okay, and also what you can do if you want to raise a question, you should also have the ability to use the raise hand feature in Zoom. Okay, next slide, please. So briefly, before moving on to our presentations, I wanted to just give you, I've got a couple of slides here that just try to give you some background by way of context to explain a little bit about the work that's been going on in EOSCUB and to just give you the context of the reason for holding this session, really. So within EOSCUB, there's some joint work currently going on actually with WorkPackage 12, which is led by Sergio Andreozzi from EGI. If I roll back, though, to nearly a year ago last summer, the EOSCUB project published proposals about the Federating Core, about its organization and contents, and I should explain that in those proposals, the Federating Core was proposed not just as the back-end non-researcher-facing elements of the EOSCUB that were essential for its functioning, but in addition to something that we called shared resources, which were services, a subset of value-added services that we felt were actually essential in order to deliver the value-add to users and to providers that the EOSCUB needs. So in the proposals that we published last year, the proposals included sustainability of the EOSCUB and we proposed that the delivery cost of the capabilities of the Federating Core shall be sustained by EOSCUB funding. In other words, the EOSCUB should sustain the costs of providing the benefits of open data policies to a wider community of users. Therefore, the EOSCUB needs to create the financial vehicle to cover the costs of the core, including provision and consumption of what we called EOSCUB shared resources beyond their originating communities. A very important corollary of that then was that the coordinated provisioning and funding of the Federating Core is expected to bring economies of scale by aligning investments from member states with the compensation of the marginal costs associated with cross-border usage of depletable resources and services. And it's this issue highlighted in the blue box at the bottom of this slide that is really the focus of the work that is now kicking off within Task 2.3 jointly with WorkPackage 12 within EOSCUB and which is the focus really of the discussion today as well. So next slide please, Iris. Next slide please. So I'll carry on talking whilst the next slide appears. So on the next slide there are actually a couple of examples of what we call shared resources and it's important to emphasize, please don't focus too much on terminology. There's obviously a little bit of a problem with the slides. I'll keep going anyway because I'm sure that will get resolved. So what I wanted to do was to give you examples of what we had called shared resources. So in other words, examples of these types of services that would provide value add, but for which I think it's necessary for us to figure out what the business model would be for their provision. So I put two examples into my slides. The first one is high performance European distributed cloud storage. Sorry, cloud storage environments for secure access, staging, downloading and deposition of large volumes of data across national, institutional and research infrastructure boundaries. The second example is high performance and high throughput distributed compute capabilities for big data processing and analysis, including simulations. So I think these examples are quite familiar to you and I think they're the sorts of services that have already got some experience. There obviously are examples of them being provided in a federated form already. So I think this is very much the territory that we need to dig some more into. So as I've said, we've got some work that's really in its early stages just now in EOSC Hub examining possible collaboration scenarios, analyzing how the EOSC might successfully support fulfillment. So what we're doing in fact is we're trying to simulate negotiations between users, so use cases and suppliers who would potentially be able to supply in response to the use cases. We're in the process of organizing workshops to try to simulate negotiations between the user side and the supplier side in order to identify what the issues would be with trying to satisfy these use cases and actually achieve fulfillment and where possible actually identify possible solutions which would help to overcome the issues that we identify. So early days at the moment we don't have any results, we haven't conducted the workshops yet, that work begins and gets underway next week, but that's the context against which this session is taking place and I just wanted to explain all that to you at the start so that you knew kind of essentially where this session is coming from. So that's it from me, thank you. Now we move on to three short presentations from our our invited presenters. The first of those is from Ilya Levinson, who's from Estonia, so I'll give you a brief introduction of Ilya while the slides are set up. So Ilya works at the University of Tarthune as a specialist in cloud solutions. At the moment he's working in the EOSC Nordic project where he leads work package three which concentrates on services and service providers and he'll explain a little more about that I think in his presentation. He also acts as an architect of the Estonian government cloud and Estonian scientific computing e-infrastructure. So I will hand on to Ilya. Hi, thank you for the introduction and opportunity. So I wanted to discuss a bit or present our experience in figuring out what it takes in order to actually start using services across the border with a lower overhead than we saw so far. Next slide, please. Just a bit of background about the EOSC Nordic projects, projects for those that are not aware of that this is essentially a regional project with a somewhat healthy representation of different stakeholders, meaning both service providers and governing bodies and the national infrastructures from all the Nordic countries. It also is a quite friendly project so we don't have much politics and we are working towards making it possible to achieve the goals of the EOSC. Next slide, please. A bit of history is that before the EOSC Nordic project started, there was a different activity funded by Nordic infrastructure collaboration or a Nike called Dillinger which ran over several years and was concentrating exactly on the same topic and that is how to enable, not even enable but how to simplify the collaboration across the borders. What are the issues and opportunities that come from the fact that services are being more open? The experience of that project was actually quite important when designing what we want to do within the EOSC Nordic and most of the partners are actually the same so we have representations from all the Nordic countries there. The results are open and you are welcome to visit the link so this is not so important at the moment but within the Nike Dillinger the scope was relatively small so we concentrated on essentially figuring out what are the issues when we try to give access to smaller scale HPC projects across the borders. Next slide, please. So basically the main outcome of the Dillinger was that whenever we were talking about technical issues they were generally easy to solve so the amount of time that went into them was quite negligible. The main issue was with aligning of the semantics and organizational aspects of the services when you want to basically offer services across the border and that is in a somewhat homogeneous environment in the Nordics. So a lot of time and discussions were spent on figuring out what are the access policies, whether they are published available, what is the level of assurance that the service provider expects to have for a particular organization, what is the impact of sharing resources across the border in terms of the additional value provided and whether it's the subject to the VAT law, what is the motivation in general beyond the project for providing such kind of access as well as reduction of the like what we need to figure out what is needed in order to reduce the amount of information that the researchers or people that willing to get access to the services would have needed to provide upfront. So we tried to avoid the submission process becoming lengthy and that was basically one of the useful outcomes. So what was clear that we would very much need a certain protocol in order to share services and ideally we saw that EOSC would be providing this kind of protocol. Next slide please. So inside EOSC Nordic the part of the project which deals with the services and service providers is essentially split into two main tasks. One task is about essentially mapping and integration of the Nordic service providers within into the EOSC and the other part which is probably not less even perhaps more important is assuring that the semantical and organizational interoperability of the services and service providers is aligned. So we're not dealing so much with the technical interoperability of the services as we're working with the let's say semantic and organizational one. Next slide please. So the reason why we want to have services on the fixed level of maturity is that then it makes it much easier to do some common assumptions or reason about the services. So it's easier to integrate with the EOSC portal. It's easier to explain to the users why this or that is needed. It's easier to make some common assumptions about the services and basically we decided to have a maturity model as a tool that would allow to at least map the service providers and provide them with suggestions as to what is needed in order to make that happen. The model is actually something that I'm going to present a bit more in details below. It will be also available later probably next week from the EOSC Nordic websites along with exactly exact outcomes for a number of services. Next slide please. Just to explain a bit where exactly we see that the maturity model fits. Now we are not working with local services on the level of providing them the motivation as to how to participate or as to why they should share the services or participate in the EOSC. Basically this is something that the services must decide on their own. What the maturity model is helping with is to align more within each country as well as across the border so inside EOSC with the common processes and policies that are expected that each service provider would follow. Next slide please. Just to give you a bit of an understanding as to what extent we went for with analysis. So we picked around 50 services of quite different categories from different countries and we ran this analysis for each of them. So for each of the services we basically have a somewhat fine grained analysis across different categories and attributes and we end up with a certain ranking of the services. Now the ranking is not intended to be a source of competition although of course it's a bit of that but it's intended to help service providers to achieve a level of maturity that is needed to at least be part of EOSC and go beyond that as well. Next slide please. So the model is designed in a way that if you achieve the minimal level of the maturity then this is exactly what is needed to be classified as an EOSC service on a minimal level. The model is intended to be simple. We know of the drawbacks that come from these simplifications of some aspects but it's also intended to be feasible to use for a number of heterogeneous services. So this is a sort of say first entry into the EOSC world. So it's something that we intend to provide to the service providers that would like to start working across the borders. So the model includes several main chapters. If you can go to the next slide. Next slide please. I'm not sure for me the slide is still the same but... Yes I think it looks like the slides are tending to stick at some point Ilya. So if you can go on you know while we try to sort it out. Sure so basically we have a somewhat traditional service management part which is dealing with the level of support for the service provisioning. However we also get into the data management and fair data requirements providing initially a set of small and simple questions to figure out if the service provider at all is aware of the fairness whether the research data cycle life cycle has been defined and things like that. We also get and this is something that was quite clear from the data project we discuss the accessibility and legal requirements in a bit more details. So we are trying to figure out if the service is intended or in general accessible by users outside its original community. If the usage from other European countries is possible at all. So GDPR analysis and limitations whether the IP rules are clarified for some of the services. So the reason why we ask this question is that they have been causing a lot of discussions and essentially wasted time. So we are trying to figure out if we can provide the answers and the best practices ahead of time. Finally we consider also sustainability of the service although this has been a very complicated question for a number of services. And also the last and the most complicated one is the compliance with the IOSC architecture. So so far majority of the services have all so to say knows in in that section primarily because the relevant documents are not yet published so it's very hard to evaluate whether the service is yeah whether the service is compliant with IOSC regulations or not. So what you see on the last slide is an example of an estimation where service providers basically saying that they have no clue as to what they need to do in order to be compliant with the IOSC. So yeah so this is basically the last slide and what I want to say is that we're going to publish this model that could be seen and used by others as well. We are running this is version two by now so this has been several iterations of that one. We plan to release a newer one at the end of the year once the IOSC working group results are published and very much hope that over time this would be essentially a deprecated model. So we will get all the required information from the IOSC itself so whether it's a sort of description template or or something else. So thank you very much I think same on time. Thank you very much Julia and I can see that there is a question in the chat actually so I think we got time for one question at the moment before we move on. So Owen Appleton has asked he said Nordic collaboration results look very positive and have yielded great results but can the model be used elsewhere? Would the collaboration be so easy between France, Germany, Italy, Belgium and so on? How do we tackle political and organizational barriers? It might be interesting to hear your sort of Nordic perspective on that perhaps before we move on. Yeah I'm not sure if I'm the correct person to present the whole Nordics. In a sense I think there is a point that the larger you become the more issues you have with the political aspects which can be corrosive if you want to achieve something. I believe that the correct way forward is essentially figuring out what are the options to increase the trust towards each other so that it would be more synergy than competition in some aspects. Like we're not talking I mean Nordic countries also compete and there is no question that everybody is providing services for free or not the only difference I see is that there is a habit of understanding other countries needs and be open about addressing them as well. So I believe that this is a complicated question and it's not technical at all. It's very diplomatic. Okay well in the interest of keeping moving forward with time I'll say thank you very much Ilya for a very interesting presentation and we'll move on now to the next presentation which is from Sai Holsinger who works for EGI. He's the Strategy and Innovation Team Lead and Business Development Manager there and he's he's got around about 15 years of experience dealing with EU funding projects and developing and implementing infrastructures also leading commercial exploitation and at the moment he's the coordinator for the EOS Digital Innovation Hub as well. So Sai I'll pass over to you. Okay I trust you can hear me well and see my screen. Yes thank you. Great so I'm probably going to come at this with a little bit of different angle which is just kind of looking at kind of pay-for-use style situations and some of the typical let's say processes and documentations and experiences and rationales that EGI has experienced over the last couple years and I think that there will be most likely some parallels with some of the conversations that will be happening inside of EOSC and I think that this could be an opportunity to kind of share some of our issues and then maybe also some opportunities in the future where we can maybe kill multiple birds with one stone. So I am I'm going to give you just only 10 minutes I'm just going to stick to the highlights here which is a brief history not of EGI as a whole but just of our pay-for-use activities what has kind of been some of our work or results to date a little bit of like a list a checklist of the types of documentation or agreements or tools that you would either need to think about or to start to put into place and then what are some of the opportunities that we see moving forward because obviously it's not a perfect situation at the moment it's just a periodic work in progress let's say. So back in 2013 basically just a couple years after the EGI foundation was set up we kind of started off with an exploratory model of saying okay we're a public funded infrastructure if we wanted to add pay-for-use services or business models into our existing or existing organization how would that work because we would have to basically make sure that we weren't compromising any state aid and competition laws etc so what we that kind of got put together drafted articulated approved by the EGI council to kind of move forward with like a kind of a proof of concept so we kind of started to pull together a number of our different providers from a number of different countries to start to kind of do a deep dive analysis on what this would actually mean within individual organizations because the complication here is is that not only are different policies they're different or vary across different countries but also in between institutes what they can and cannot do so we had to start to find out what are these combinations of who can do what and under what circumstances so one of it was from a policy point of view the other one was from a tool point of view so could our existing tools even basically account for the usage of those services on a pay-for-use and be able to calculate out a price and and eventually a bill um so then we needed to look at some of our other agreements like having service level sla service level agreements between between the different parties operational level agreements is something that you need to think about in a federated environment so you have your back end providers all coordinated and then we kind of started with okay we have a baseline documentation we started to then come up with let's say some requests or some use cases some outside of projects some inside of projects and i think that's what basically brings us to the kind of variable scenarios by which different models need to be put in place under a different situation but the the big question that we were really trying to answer was after we had let's say one single central project that was funding all of not just the coordination aspects but the physical infrastructure once you move away from some large central project and you move into multiple projects and multiple support therefore you have to have multiple models in place and each of these come with different pros and cons so just in participating in horizon 2020 projects you can find that you either can participate as a beneficiary which has pros and cons associated to that you get visibility local government recognition but whenever you're coming inside of a project and you think yeah i want to bring in an infrastructure that has 250 data centers spread across 40 countries you blow up their consortium list and they're not super excited about that so then you think okay well what about using the third party mechanism this is interesting because basically it reduces the number of beneficiaries but then your individual providers also start to lose a little bit of visibility this also forces the projects to already know their existing technical requirements in order for you to be able to do provider selection so then we've run into models where basically they don't know where the requirements are they just allocate a side budget and say you manage this budget and you identify the providers whenever the requirements come in so now this is a completely new mechanism that we have to have in place but it is very very similar to the types of things that you have to do in order to basically run a pay-for-use model in addition to that you have external requests that are coming from outside of projects so either EGI is just purely a matchmaker so there's a single user in a single country it doesn't make sense for EGI foundation to play the middleman in that scenario so we match make them to an individual provider in that country but for other cases where you're basically having multiple providers across multiple countries without having some type of central broker in place then you basically force people to have individual agreements with individual country and that doesn't really uh this doesn't really scale so in that scenario what are our pricing models can can the individual providers provide you the pricing what would be any type of broker fee that we would put on top of it etc etc so this is basically what we've been trying to use we've had some opportunities in EC funded projects where we can kind of refine this mechanism so in the next JOS project there was between 10 up to 12 pilots now and this is the scenario where just EGI was allocated some external budget we would collect the requirements from the individual pilots do the provider selection and then um and then allocate the resources so you still have to have agreements in place the providers still need to be able to invoice EGI in order for that to then be an eligible cost with inside of the project so we were allocated 100k to basically redistribute so far we've done us we've allocated 60 000 of that to six EGI providers um and something similar to inside of the EOS Cub project through the digital innovation hub here um we were restricted to only being able to redistribute any type of funds to consortium members but again the model is still the same collecting requirements putting agreements into place making sure that those providers have provided you with the cost that you can account for that then you can put inside of your budget and that gets associated to so this these two projects were quite nice because it started to kind of help us really kind of dig down into the processes and the procedures and the documentation this also led to we had an external request um through the the DS and the European Space Agency Xprevia was the company that was let's say the lead contractor but here the EGI was contracted out to do a kind of a proof of concept the validation of some some aspects with inside of the DS if it was able to run in multiple providers in multiple countries so this was a very small contract that was a proof of concept that ran over 12 months but it ended up being quite successful and now we've just gotten requested to continue that second contract the other thing that people can start to think about is just basically monetizing individual services so for an example we have the FITSM and ISO 27 000 trainings that we offer as in-house courses or open registration courses which comes with a specific pricing scheme and a specific pricing model so start to look at how can you monetize existing services that you have and then make those available to the rest of your community so in just terms of documentation in the last two minutes that I have so we had to basically collect these kind of what we considered letters of intent so what we needed was a clear statement on the ability and willingness for the individual providers to to provide pay for use services so it was their declaration that they were able to participate in such a model and then also under what conditions would they provide these paid for use services because some of them have to offer services free within the inside of their country but then they don't necessarily have to outside of their country and every single country kind of had their own different policies so we got that into a statement um you have to be able to collect these service offers from the individual um from the individual providers um right now we're tracking this in a google spreadsheet it's becoming more and more sophisticated over time but is this really a long-term solution probably not now virtual access mechanisms are now being requested for example in these eos projects so this gets into how do you what is your unit cost what are the units how do you calculate that cost so we've come up with some templates for our providers to start filling them in but this isn't let's say as easy as of as it is of just having a template and filling it out so some support is needed and some additional maturity in that we need to be a little bit about the selection criteria so it really can't it isn't always about who's just the cheapest provider a lot of times it needs to be better with what is the exact requirements that are needed so some have higher availability and reliability targets versus others um some just had very specific technical uh requirements by which are maybe only a subset of our providers can fill you do have to have an sla so for an example with egis the foundation is the broker we have the sla with the end customer but for each one of our providers on the back end whether it's in italy spain france we have to have an ola on the back end to make sure that our agreements are aligned and then we add some type of broker or coordination fee on top of that so some opportunities moving forward is basically just having a pricing tool increasing our automations we had some stuff in our egi accounting portal but we are looking at how can we have a better collection of the individual pricing from our individual providers so we're in discussion with some swiss guys and this is basically my last slide which is kind of the lessons learned so we acknowledge that funding sources is like an issue and we need to basically create diverse opportunities for us to be able to redistribute funds to the egi providers so the horizon 2020 projects are helping with that um we're getting more requests coming in so we need to kind of increase the automation of it so the overhead reducing the overhead of the agreements um trying to have clearer pricing listing and procedures um right now we only have about six providers who are taking full advantage of the pay for use model so we need to kind of understand how can we scale that out but right now not everybody is able to provide invoices etc so we just kind of want to continue to investigate that and we are now investigating other services that could have a pricing model so for an example we're in conversations with two providers around for example archive storage and stuff it's just that since egi is a public funded organization we're a non-for-profit we always have to kind of walk this uh with this boundary to avoid that vat charges end up on our council fees so part of our long-term activities is looking at a potential commercial arm if it becomes interesting so just a quick little highlight about uh egi and our current activities and the pay for use and with that i'll stop and then take any uh any questions that you may have thanks very much sigh that was really interesting um so i can see that renny has asked a question which i really don't think it's possible to give a short answer to so i'm going to avoid taking that one now because i think it will come up again during the the panel discussion but um bob jones has asked um in the pay for use model what is the legal basis of the service level service level agreement what happens in case of a breach of the sla such as non-availability of the service and how is dispute resolution handled can you make any sort of relatively brief comments on that uh so basically it's described in the sla so every single aspect that he covered is described inside of the sla it would take a lot to go through every every bit of that but um also some customers have different requirements in terms of basically so some i'll just give you one example where we had to say no is we had um with the european space agency to do a direct contract with them you would have to be basically you are fully liable for the exact amount of the of the contract so the bigger the contract the more in which you're liable which then you need to make a decision over whether or not you are uh able to accept that number of financial risk and at one point for one of the large contracts we just basically had to say no for others it's um um it's more lightweight depending on how it is but it's all described with inside of the with inside of the sla so that's clear for both parties before and if the individual customer uh isn't able and willing to accept under those conditions then you just have to say you just have to say no so i think one thing is an important message for everybody in this environment is it's also important to understand your own limitations as well and not just kind of go into any agreement uh any agreement blindly so um yeah yeah excellent thank you very much okay so i mean there are some really interesting questions in slido um actually but i think just in in the interest of time i think we need to keep moving forward so thank you very much to sigh um i'm going to introduce um our third presenter who is mark van de sanden um who um is a technical specialist at surf very experienced in managing large scale computing and data infrastructures he's also the eu dat technical coordinator as many of you i'm sure are aware um active in many eosk related activities and is the eosk hub representative on the eosk architecture working group as well so i'll hand over to you mark thank you okay thank you um not sure if you see this slide or you see the presentation mode we've got presentation okay how can i switch swap displays okay yep that's good okay uh thank you thank you for the invitation uh within this uh this presentation i want to present a few of the examples in which eosk has been working to collaborating with a number of communities which are already many much uh distributed across many different countries uh and also making use of different service providers within uh within eosk if you look at that eosk what is eosk providing providing in general uh building blocks towards research communities who can uh building up their own research infrastructures on basis of standardized uh building blocks and if you look at this slide you see that we have different types of building blocks which can be used for different kind of different kind of purposes eosk that is an organization this started us from from two different projects but there's currently a legal organization and a partnership agreement which has 20 different organizations participating across 15 countries within within europe um if you look at the different services i'm a little bit annoyed by the diagram uh if you look at the services uh we provide different services and if you look at how the services are being accessed uh the users are coming from 110 different countries uh and which is spread across the 14 different installations which we provide and if you look at the data which you make available it will be defined this house is big across 24 different communities from different repositories so in this case you can already see that we provide different kinds of services and but we have to provide different services across uh different borders and this is a few uh of the uh cross border use cases which we have this is one of uh c data net uh we participated with them within the c data cloud in which we have built up with them uh and cloud infrastructure on basis of uh the dot services which is spread across five different uh uh partners within you that but also five different countries within europe but if you look at the data net as an organization there's already an organization which is uh spread across more across europe where 110 different data centers work with them to do the research and analysis so you can see that there's there's a very highly distributed infrastructure but also highly diffused uh different kinds of user domains so they are not coming from a few countries but really spread across many different countries uh within icos is uh in a different kind of use case where we uh participate with two different partners within uh uh within u dot but providing services towards icos as the larger uh organization which again uh is already in european organizations spread across europe providing different services collecting information for many different stations across europe but if you look at this we set up a smaller uh uh use case where uh one of the sites mostly acts as one of the processing sites and another uh partner was in u dot uh acts as an archiving site where the data is then replicated for for safekeeping uh the combined that use case if you look at this uh use case uh they uh generate the data at one of one of the sites which is a shitnet run uh but they have the uh users uh located at many different organizations also having access to different kind of uh data and computing services where they want to do their analysis for be able to do this we collaborated with them we set up and be the safe application environment where the data is ingested in one site the data is automatically replicated at uh two different wallet sites where the users are being uh giving access to the data where they can do the processing analysis of the data but also again can publish uh the data on all the service but if you look at this this again that the data is at this time generated at one location but the users because user base is distributed across many different uh uh countries the data has to be replicated across those countries to provide uh access uh to the data but also in combination with computing resources which are required for the simulations processing of the data it was very similar use case um then i want to go to to the challenges which we have seen if you look at that uh in that is is a service organization built out of generic and traumatic service providers which are spread across across europe that if you uh look at the service providers they have different mandates one is the mandate which are given by uh the national organization because you are a public funded organization and in the other case the mandate is provided by uh their community and that will set uh some limitations in providing what kind of services can you provide or what is the amount of services you can provide uh to uh to the users outside of your own uh mandate and we have seen that providing services outside of your mandate can be fairly limiting to a long number of percentages and if you look at eosk it is mainly about providing services outside of your outside of your mandate so how do you enable this how can you provide those services but also how can you recover those services the the recover the services and the resources provided to outside of your mandate one of the other challenges i think that is also what i already mentioned all service providers have different kinds of models in uh developing the cost prices or the prices and we did some exercise on this to see how can you come to a similar cost model and that this seems to be very challenging but also even if you have the same cost model it does not mean that it will be the same cost or the same price and also to prevent that you always go to the cheapest service provider you need all the kinds of mechanism to select who is the best service provider for providing services and resources to uh to the customer and the virtual access model that is already explored within uh eosk hub um it was not for me not the the best model that seems to be that it has been approved by differentiating uh some of the regulations in 2019 so the model which is an n5 eosk hub 7 is very improved but it's still not not perfect one of the challenges which i see within eosk is uh the virtual access model is limited it's only limited to all the services services instances which are listed in the project proposals and it is not open to all services which are currently listed in the eosk marketplace so if you have onboarded your service into the marketplace then you still cannot choose all the virtual access model because you have to be included one of the project proposals um it is also not really optimal for uh a cost recovery of services where uh the users can uh register themselves or even you do not have to register those kind of services because the virtual access model is about okay you should be able to demonstrate which is the users beyond your local local community or your local customers how do you want to detect new users for this kind of service and how is the new users while maybe other users are left and do not use make make do not make use of the services again so there you need different kind of uh detection of what is a new user and how can you show that you have new usage another challenge which i also see within uh virtual access is how do you do cost recovery of core services organized by e-infrastructures for example with ignore that we need a number of core services to operate as an e-infrastructure and it's very challenging to get cost recovery via virtual access on those kinds of services to mitigate one of those those challenges what we always try to do is to find natural relationships between the customer or the request of the service and the service provider so it is more a brokering function which we provide with a new dot we have seen this was in uh the cdata cloud project was in we see this was in combined let's we see that this enables and i think therefore it is also a natural match because then you can also circumvent that the service providers are more closely related to the customer with the community therefore they are easily to provide services resources for those customers um and if we select uh service providers for providing the services then uh the slices mostly based on affiliations with the community but also is a direct relation between the request or the service provider or the community across the countries where services are we provided okay thank you if there are any questions uh another answer them thanks very much mark that was really interesting i think we've got time to fit in one question briefly um so oan has said in in slido um are eudat cases paid or are they part of project participation projects proxy for the complex rules agreements mandates needed for cross border use collaboration in projects is great but i'm not sure it leads easily to potentially paid cross border resource provision so perhaps you can comment a little bit in response to that please uh it is it is both ways uh if you look at the cdata cloud uh that has been uh built up to uh uh within the cdata cloud project but the continuation of providing the services is going beyond the project because you do not build up an infrastructure services just for the project if you look at the icos use case that is a paid service contract made through eudat where uh cc and eudat are service providers so we have an asla with with icos as organization and overlays between the service providers and eudat limited which is the contract owner of of this contract okay thanks i actually see um i hadn't spotted this before but there's there's a comment um from janus mahaak see in the the zoom chat how do you explain the costs if eos principle says free at the point of use is that something you can comment on mark um hi it strongly depends on who who's going to pay or where this cost for economy uh eos is is is is focusing on a free at the point of use but uh services are not free resources are not free uh somewhere the cost should be recovered and i think there uh uh within the infireus go seven you have the virtual access to do the cost recovery so you can provide services and resources for free but somewhere the service providers are being paid uh but i said uh within the infireus go seven this is limited to uh the service instances provided within the project proposals there's not an open model so i see challenges for providing large amount of uh resources for free in general for eos if there is no cost recovery possible yeah okay thank you and i think that that leads us probably quite nicely into the panel session anyway so thank you very much mark that was very interesting and uh so we've got half an hour left now uh for a panel um and you should be able to see now on your screens that we've um we've moved or we are moving into um a poll session on slido so um i've got six panelists here who i'll introduce briefly and we've got some questions that um will be asked to them but in parallel with that you will be able to see each question as well so you can see the first one now released as a poll in slido so please feel free to um add your own comments in response to that um i'll now introduce the panelists briefly um which will be an alphabetical order since of course they're not actually sitting in a row um so first panelist is Owen Appleton who's the service portfolio manager at EGI um and also deals with the same sort of subject matter in the yaskub project um he's been working for a long time in the infrastructure in previously in EGI projects um whilst at CERN um he has a background in public relations he's worked as a consultant on policy and exploitation and he set up a national e-science support network in switzerland um he was also a co-author of fitism um that's our first panelist second René Belceau um who is from Denmark he's got more than 20 years of experience with research policy and research infrastructure provisioning primarily um in the area of ICT um i should say i i can't necessarily see whether you have or have not switched your video on but if you are able to switch your video on panelists then please do so um so René has worked in national and international government political institutions he's been the head of the Danish Center for Scientific Computing and he's now working for the national e-infrastructure provider organization um with a focus mainly on international issues around HPC and data management um and he's the Danish member in the yask rules of participation working group currently um thirdly we have Lena Munari I'd like to welcome Lena from the European Commission um she's the deputy head of Unit C1 in DG Connect which is the e-infrastructures and science cloud unit um based in Luxembourg she's been in the EU institutions for 20 years much of which has been spent in DG Connect dealing with policy areas including cultural heritage, digital libraries, digital preservation, technology enhanced learning, fourth panelist is Antiporsula from Finland who's the development manager for international data services at CSC um he's very experienced in IT infrastructure development operations and policies he's currently co-leading work package two of the yask hub which is strategy and business development and um his current role in CSC includes responsibility for development and operations of EU data services and managing a portfolio of yask related project participants so welcome Antip. Um fifth panelist is Steve Robert Shaw um who's had an involvement in EU research projects since the fourth framework program um and in national and bilateral international programs for even longer than that um he's got a wide experience so in a sense brings a wider perspective to this panel um dealing with telecoms, distributed computing, AI, web services and data for many many years um with experience in academia, government and industry in a variety of different roles um he has involvement in the yask through a number of small contracts um but is familiar with EGI and EUDAT services as well as um some of the research infrastructure user communities um and currently he does have a focus on sustainability so there's there's quite some relevance in terms of his recent thinking uh to this panel. Um fifth but sorry sixthly certainly not least is Federica Tanlongo um from Gaer in Italy who coordinates the communications and external relations service for Gaer um she's been with Gaer since 2004 um and has had involvement in a number of different EU projects currently she's creating um she's working to create the Italian computing and data initiative and also in the EOS pillar project um so EOS pillar covers Austria, Belgium, France, Italy and Germany one of the um five B projects um so in that project she's the deputy project manager and she's the leader of their work package four which is from national initiatives to transnational services so I think some very good panelists with um some very relevant and varied experience to offer us so um without further ado I'll enter into the panel so the first is not actually one question there's actually two or three questions rolled into one here I hope you can all see it on your screen and I think I'll just go through an alphabetical order really for for the first round at least um what is the incentive for public funders to provide national services across borders what's the incentive for public service providers to do cross border provision of resources or services and what mandate do they have for this so Owen if I can turn to you first please sure um I think for the for the first part the answer has to be reciprocity in some ways um for funders to support um cross border provision of services something has to be coming back but I think how how that comes back is variable it could be that money flows out of a country but it also flows back in and there's some sort of balancing of that as well like the balance of trade of services to make sure the countries aren't disadvantaged or alternatively it can just be based on um reciprocity in terms of mutual benefit so making sure that if you are essentially subsidizing research in other countries in some way other countries are subsidizing your research through providing new services that you don't have locally that's obviously very difficult but I think that's the only reason that the funding agencies will support it and as to the other part what's the incentive for the providers it kind of comes down comes down to the same thing I think they have to feel like they're getting something out of it one problem that I've definitely seen uh in previous positions is that even if there are parts of say a research organization a university that are very keen to support the wider research community because they feel engaged with it they don't necessarily have the organizational mandate to do so outside of as I suggested in the question to mark a project structure which is always proxied for those mandates and I think that uh there there is work at a at a high level at a policy level at a funder level to make sure that the research organizations have some sort of economic model that supports cross-border provision of services and that they then make sure that their organizations the people within them have the mandate or feel they have the mandate to do that outside of just doing uh doing a project I know it's much easier to do a project than trying to do a public procurement for instance in terms of the reaction you would have from your administration I'll stop there for now thanks Owen okay I'll move on to to Rene well I think the problem is basically that there aren't any incentives or at least they're extremely weak so for for a public provider a national infrastructure provider to send its funding the funding it has been given by the ministry procure procure in one of the other european countries basically be to undermine its own services and its own staff's income so it would be pretty much like asking in the old days uh in the industry days where we had like shipping to ask a shipping country like Denmark to build its ships in in uh in Germany it just will not happen unless Denmark and Germany on a governmental level agree for some level playing fields and agrees to open up the market and make this possible so so the the answer is there isn't any incentive and in terms of the other way around uh offering something well then is the cost recovery and since there is no business model there is no possibility to cost recover it will not happen and when it does happen it only happens on the experimental level where it's interesting to maybe share some some knowledge maybe expand your network and so on so uh when it does happen that there is a cross-border sharing uh it's of extremely marginal extent at well pretty much you can't really say it's happening so it's a business model that's completely lacking yeah yeah okay thanks Rene okay i'll move on to lina now lina could you comment please is lina there yes okay sorry i have an issue because my can you hear me yes thank you yes because my screen is frozen actually i can i can all the time see all the participants now the humble number of 140 so it actually takes also the one third of my screen so switch on my microphone but anyway yes i mean these are interesting questions obviously from the commission's perspective we're looking at you know the the virtual access unit cost model has been mentioned we all know it's perfect uh obviously from the the commission perspective i think what vene said and for oan as well i think both these cases are valid i mean i think the reciprocity is part of the business model and unless there is a business model there will not be any reciprocity so i think from our perspective EU clearly funds only what is subsidiary to the EU cross-border use so it is not our task or our interest in funding in big chunks those services that are actually funded or international level but clearly as the EU is all about free flow of everything if the service provision is not able to flow outside of those silos that are currently in place because of the national structure and there's obviously nobody's fault because he's the way we've been installed we are not going anywhere so i i note also what what we say about the experimental use that's very true i don't think we're anywhere closest to a proper business model and incentives for sharing and i think this is something that needs to address very quickly but obviously commission is not necessarily the one who is at the central place of this because i think the initiative has to happen between them in the space between the major players and to start building major use cases where this is actually needed because you need to demonstrate what is the actual need because you don't just build things for the sake of building them but but this just also obviously not not undermining the fact that that that that the whole of the cross-border use is extremely important because otherwise we can just go back into our own national silos our research silos and and and live happily ever after so this is my comment for what is yeah thank you lina and i see there are a couple of good comments in the the chat in slaydol please keep them coming i'll move to auntie now thank you can you hear me yes yes we can thank you yeah i can think of uh several motivations for or incentives for providing services across borders i can start from the service provider point of view so i'm i'm representing eudat and also the csc center in finland so i mean the users the customers they want services for research is international the research communities have been going together for a long time to form european research infrastructures etc and and those really need services that are also european international in that sense so there you already have one incentive for service providers to answer to this demand also the researchers they want services that are competitive and they are good services they people are used to commercial cloud providers and and so on so i don't think it is for researchers for researchers so important if the service comes from the neighboring cloud provider or from another country but of course then the business model models is the different thing we come to that and i think also for the research funder we see that that even if the focus is national i mean national funders they also see i think in many cases that you need to really support the international research for your country to be competitive in research they need to have good tools for collaborating which means interoperability and services across countries also if i may add i think it is a motivation to to be part of the european ecosystem and and really support the european research i mean we should we should be european here and support each other yes thank you that's a good point um i'll move next to steve robert shot if you could comment please steve yes hello um i apologize first of all that you can't see me i'm in the uk on a rural internet connection which means i very rarely i'm able to supply video streams to anybody okay well at least you're with us in voice at least i'm with you in spirit and voice yes um so i i think the i i'm i'm going to slightly turn this question on its head in order to answer it i hope you don't mind so the first thing i want to say and it's been touched on already is that one of the reasons that the commercial cloud providers succeed is because the users trust that it's always going to be there when they need it even when occasionally it's not so one of the reasons in my understanding that um the e infrastructures initially and now eosk is being considered as a pan european service provider to european scientists um and the models that we are considering at the moment is to challenge that attitude amongst users that these facilities if we can call them that are project based and therefore they can't be trusted to always be available even to the scientific community when they're required so by providing something that is able to build trust then users will want to use it and therefore they will put pressure on their national authorities and that is a form of incentive if you like coming from the ground up the problem with that view is that it doesn't take into account the fact that the um the the public funding comes with a political element attached to it and we can't always rely on politics to play fair and have the same kind of high-minded moral attitudes to european collaboration that we all would agree is a good thing we have to consider that it has already been demonstrated that some nations are quite capable of withdrawing funding not only from their own national contributions to international efforts but also in wanting to actually withdraw from the international effort themselves so it comes back to this trust so the incentive for the national contributions has to be based on building trust not on collaborating committing resources or collaborating um through high-minded ideals we've got to get some kind of concrete agreements at a high enough level to build the trust and maintain the trust in the services being available whenever the users need them thank you very much very interesting point um and finally for this question i'll turn to Federica i am thank you so um okay many of my colleagues stressed the point of the limitations given by the the fact that some funders are public and they come with this political part that touches that's of course the major problem in my opinion to deal with here and because and it also reflects on the second part of the question the one about service providers because if they are publicly funded these these against to the to this level as well so we in pilar we asked this question to ourselves and we it was part of the survey we did in the country so we discovered that many publicly funded services do not have at the moment access restrictions so which also includes transnational usage so in a sense these kind of services and we are talking about in particular of service provided by large research infrastructures are already free in point of views but this can be either seen as a byproduct of this the main activity of funding the national level or as an investment as Owen at the beginning was was saying so you you are a political organization you invest where you see that you gain more than you spend so I believe that while the residual activity can only work to a point and there may be differences in terms of the percentage a country can allow for this residual resources and even if at the stage uh those resources are uh let's say overabundant when they use uh we go to full speed sorry when the resources will go for speed the they turn out not to be enough especially if we are considering the long tail of science and education and industry and let alone also the the citizen scientists as likely candidates to use these resources so to me the only valid incentive to scale up is political willingness to be part and an active one of this wider picture and to understand that if everybody put something on the table then there will be so much more resources for everybody else I think this is something you already see in the large scientific collaborations and it doesn't come for free because it came it came from long-standing negotiations thank you very much Federica interesting observations so I see there are some comments in the the panel in slido thank you keep them coming they're very interesting I think because we're getting quite short of time what I'm going to suggest is that we move to the next question the next which will be the next poll and rather than just go routinely through the panel members individually what I'd like to suggest is that the panelists simply indicate if they would like to comment but also I'd like to encourage the audience if they would like to speak so audience members you can use the raise your hand function in the zoom chat that's not something that's available to all the panelists and presenters because you're all co-hosts so you would just have to speak out verbally if you wish to speak so if anyone would like to to start off so the the question here is how should funding flow to providers for them to make their services available across borders so who'd like to kick off then I could I could say that the possibility for that for national protectionism has to be removed I mean it's no different than other european issues that we have had the last 50 years so as long as you can protect your your your home service and there is no business model across borders it just won't happen absolutely agree yeah so I mean it's like if we keep avoiding to getting into the to the actual way to move the money we just won't get nowhere thanks Rene okay was that Steve who commented did you want to say anything yes yes I want to support that point of view and also we have to consider what the likely impact is of the current situation that we're all living through we could all remember the financial crisis that took place 10 or 12 years ago and the impact that that had on various economies and we have to ask questions about how willing political entities are going to be to carry on funding various international activities when we are really up against the the budget limits on being unable to fund national activities thanks Steve so if other panelists want to comment in the meantime I can see Sean that you're commenting in the chat if you would rather just pose your question verbally just raise your hand if you'd like to but um I'll stick with the panelists for the moment would any of the other panelists like to to continue with this can I say something I think one of the issues here is you know somebody mentioned in the presentation about the cumbersome way of when you're trying to deal with this through projects that there's you know all these social agreements that are already in place but you know they have to be put in so my question here is that if you want this to be supported at the UK level somebody already mentioned the broker infrastructure so I would like to hear from the panelists as well that would that be something I mean clearly you need to see which are the services available and why would they be why would somebody sort of want to provide that and what is the added value of providing at the border and then probably you know with broker infrastructure have some sort of a credit mechanism that you could actually apply I don't know these are just ideas but I mean obviously from the commission point of view is can be coming more and more important particularly because funding AOSC and AOSC related activities from individual projects is sort of coming a little bit to the end of its day I think because you cannot car you know for forever keep on supporting infrastructure from individual projects but if anybody has good ideas of how this could be also channeled through the the new projects what is the broker infrastructure something like this so I'll be very very happy to hear thanks Lena and I see there's a comment here in Slido labeling services just as free at the point of use is not sufficient and also either allow the money to follow the scientists or at least allow funding to scale with active users I would very much support that that sounds like that's what needs to be done okay just a comment there if it's okay Renae I agree but there are complexities to having a token mechanism one of the main ones is maintaining your service ready to use if you're unsure of what's the volume of use you're going to generate and therefore funding you've been and you need to be able to have some baseline certainty that you can keep the service there and justifying keeping resources particularly if we're not just talking about machines if you're talking about human expertise to make those machines useful how do I have to keep basically idle consultants there ready to serve an unknown pool an unknown demand pool I think there are some ways we could deal with this one of them is by making it very easy to purchase these resources probably like you say maybe with a token I think one idea might be just to say that for certain classes of research project perhaps nationally as well as a European scale some small percentage of your budget can be used to purchase EOSC approved services that are in some approved register without having to change your budget without having to justify it or go through a procurement because you can't always foresee the use you're going to have but if you knew that on your ERC grant or your Marie Curie grant you could spend four percent on these services and you could really only spend it on that you'd be looking around for the good services to use to support you but I think you need some semi-artificial means to maintain enough volume through this system to keep the services there otherwise everything other than bare metal I think wouldn't really wouldn't survive thanks Owen I think I think Antti wanted to comment I saw did you still want to speak Antti I think Owen actually covered this I just wanted to bring into the discussion of this kind of a mutual benefit for national funders that indeed some kind of a procuring system or even a joint pot of funding for international usage might come into question that if the national funder can kind of be assured that they will get the same amount of benefit from other funders for their users I think the Nordic countries have tried something like that as was presented in the first presentation today this I think could be I mean one at least in theory one option some kind of a joint pot of funding I don't know okay thank you and we're getting very short of time so I suggest we move on to the next question can I make a small very small comment I think that well for one thing the money following the user side the user side shouldn't be beyond the the only criteria because you may have some disciplines which are sort of nice but still very useful and very rewarding in terms of results so the size of your audience can't be the only criteria but could be one and the other thing is that in my opinion we should also take in consideration for the funding that that are not actually money and work on harmonizing them because at this stage what we see is that there is quite a lot of things that are funded through in-kind mechanisms in the countries you can see this again in the research infrastructures but then you have to agree on how to define this in this in-kind and I think it is really important because there are a lot of countries and we do count in different ways thank you okay so we're nearly out of time I think we we can overrun a little bit I've been given permission so those of you who wish to stay please please feel free to do so because actually the next session doesn't start for another half an hour or so so we'll spend a few more minutes with your agreement silently so the the next question is what is the overhead on cross-border provision or consumption of resources in terms of procurement rule contracts SLAs and so on and so forth so where are you know where are the complications and costs coming in in terms of time and effort and money so I think I'll suggest that we we've run this the same way as the previous ones so just panellists if you would like to comment please just say so and anyone in the audience who wishes to contribute verbally just just raise your hand in the zoom panel and of course you can type your comments into Slido so panellists who would like to start with this one I can I can start so I'm here so I just starting to discuss and so one thing is that at the moment that you will make a contract with your user then then you have this overhead I don't really see that there is so much difference if the user is from from your local user community or from outside as soon as you need to go into the contracting process there is an overhead okay maybe some technical difficulties in in signing papers or something like that but the principle it is the same you need to have SLAs and you need to have definitions of your service and then another point is that the service providers should also make the provision of services easy in this way or contracting for services if you are thinking of paper paper use but I think we heard examples from EU that and also EGI that how how this can be done with the central representative so that the services can be contracted from one organization instead of different nodes of a federation. Thanks Enti. Okay anyone else want to comment? Yeah I would agree there is no extra overhead it doesn't matter if the users from one country or the other of course it matters if there are more users there are maybe more costs maybe not but where the user comes from it doesn't matter as long as there is a payment for the service. Well if I don't have the hand raised function because I was a co-host if I could jump in here for two seconds yeah so there's two points here one is procurement is very very different than just having a contract or an SLA I mean SLAs you just come up with a template that you can get agreed between the two people and off you go where procurement can get very very complicated and we already starting to see that with with Oak ray and a number of different frameworks that are that are being discussed so I would separate I would separate out the two and the second point about SLAs and contracts and stuff in OLA's and all that kind of stuff actually is we are finding to be there is quite a bit of overhead when you start to scale out so for an example if you have 10 or 12 different pilots with inside of a project that need 10 or 12 different SLAs and then you have four or five different providers on the back end that you need to have OLA's with you find yourself not being able to keep up with the amount of requests that are coming in and stuff so you do need to like let's say improve the automation and to in order to reduce the overhead which is it's always just opening up a word document just to replace some key words across the template and closing and all that kind of stuff takes way more hours than people can appreciate so I think when you get in the scalability you will need some type of let's say technical service or solution to help you be able to deal with that but if you're doing one every six months yeah of course there's there's little overhead so that was my two points thanks Si I see that Paul Rose from Géant has raised his hand Paul please go ahead thanks Dal hello everyone I suppose first of all it's a bit of a sales pitch in the first instance that tomorrow's part of the ESCUB session we've got a slot there on business models and procurement where there'd be quite a bit of discussion around this point so rather than take up time now when we're very time constrained I just encourage attendees if they're particularly interested in this point to perhaps come along tomorrow and hear what myself and colleagues who've looked into this matter they're going to cover and present but of course you've got two sides to that you've got both the the selling activity if you're a public body are you actually entitled capable of selling what systems enterprises do you have to support perhaps complexity of billing and invoicing metering of services depending how you intend to recover your costs and also what liabilities has been touched upon earlier and then I think as people have already raised here as you move into a more comprehensive and mission critical consumption of resources in order to conduct your science then the more diligent you probably want to be in your buying process whether that's dictated by legal obligations to carry out a full OJU procurement or having certainty about data ownership rights ensuring all of the fair principles are in place SLAs are there to ensure that you can deliver your mission that's quite an undertaking itself so there's certainly some opportunities in the EOS model in the future as if I thought it's been positioned in some of the working papers of the sustainability working group is this concept of EOSC exchange if that center of expertise role can exist to help soften or remove that burden for many and also I'm part if funding could be corralled to deliver on the ambition of free at the point of use there's a lot of things there many of those triggers will start a whole conversation themselves let us let us tell you a bit more about those tomorrow but of course this the EOS project is a is a is a big monster and we have to take bites of this in many different places but uh yeah that they're very significant considerations yep thanks Paul good comments thank you anyone else want to make some more comments I have a comment I think the interesting thing about the question here is what's the overhead and I there are a few different things that determine that overhead some of which Paul had mentioned VAT which is mentioned in the slide there was an interesting one but I think also you have to consider how many parties are there to the agreements so if it's bilateral these things aren't as Renee said fundamentally any different to something that's within one country when they're multilateral they become much more complicated whatever countries are involved and that's when you really benefit from having some central broker or framework contract that allows you to minimize that overhead and related to that I think the overhead is also related to how big the purchase or the use of resources is and it may be that we find that cross-border procurement cross-border provision only works for very large-scale sales or purchases because then the the relatively more fixed overhead is kind of diluted among them the sad thing is that that would then discourage smaller more niche services which might include technology and infrastructure and human efforts where the each sale may be relatively small which makes the overhead on the mean credit behind because they still require an SLA maybe not a procurement but they require contracts to be agreed I need to be moved around and it's like that so I think we have to think about what kind of services we want to support and I hope that it's not just large-scale procurement of say bulk computing resources because I think there is more than that that we can benefit from this community thanks very much Owen okay yep thanks Federica please go yeah another point I'd like to make is that in some countries in order to be eligible to be bought by public institutions your service must be enrolled in some kind of or accredited in some kind of marketplace public marketplace so in for instance in Italy there are some limitations for universities or other institutions to just go and buy other services outside this marketplace so if this is the case this is part of the overhead because you should get accredited in multiple countries and I don't know how many countries do for synthies but this could be huge if everyone has something like that so if you want to provide a service transnational the potentially you have you have to get accredited in order to separate countries of the station so maybe also considering a european level accreditation is something we should think of I think it's clear that EOSC is in a position where it's starting to have to transition between thinking about itself as a project or a series of projects and starting to consider itself a business none of the problems that we've discussed so far are unusual or unsolvable in a business context overheads are overheads there are reality I think what what what's starting to emerge is that from my point of view in terms of project funding some kind of funding is still going to be required and maybe it needs to be thought of in the same way that the business would consider investment so the investment is put in in order to help EOSC build up its portfolio or services and support them don't forget that amazon ran for over a decade at a considerable loss before it started making any money at all and now it's you know it's close to eating the world so there is one of the earlier questions was what is the incentive excuse me well the incentive is through investment in real services that are demanded by real users there will eventually be some mechanism that is able to support itself and possibly even return some of the investment through some mechanism we haven't discussed yet but it's not none of the problems that we're discussing are unsolvable in a business context so maybe the the solution that we're looking for is to start thinking of EOSC as a business or in a business-wide manner thank you Steve right we're getting very short of time but i can see lina you've got your microphone open was it because you want yes sorry i have it on all the time because i told her that you know otherwise i could switch it off off again normally so it is on but i tried to be quiet sorry i have to keep it all because i'm not sure if i can switch it on okay um well i can give the last word to you just now if you want to make a brief comment yeah no i think this is all very interesting consideration i think what the previous speaker just said about the EOSC as a business okay i think it's uh i was thinking of the same thing i mean if you look at the the current platforms you know they they have been able to solve all these problems you know i live in luxembourg which is the e-commerce hub of of europe because the vat is the lowest here it's very simple to see why all the all the world companies that operate in european soil are providing the e-commerce services are buying all the procurement that's all in luxembourg but this is uh but this is a country specific thing for luxembourg i think in the same way maybe we should think in EOSC in terms of of specialization like who does what better do we have who but who does what and who does it best do we have an overview of this uh do we know which services are best provided in a certain way but there are certain entities and and maybe there is a case then to showcase like why do we all produce this if there is somebody who can do it better and cheaper and quicker i can specialize on this this is one point then another thing i was thinking is the is uh in general uh when we when we look at EOSC and its and its um development i think one of the speakers said it has said it already in the very early in this discussion is that uh when people say that they something is not possible it usually means that there's no political will to do so so i think we have to actually be very realistic as well uh the choices of countries and the choices of of different kind of mechanisms that are a bit in the important place are not often based only on rational choice so that's why uh the scenario that i was advocating might not be possible but i think uh we must be striving somewhere closer to transparency so that it becomes evident that it's not worth providing the same service in 27 times or 25 times whatever the number would be and uh and and start start to think in this way i mean uh but obviously there are all kind of political hurdles that before we get there but if if we don't take this as an objective it will also happen okay thank you very much i think we're going to have to wrap up there um we don't want everyone else to eat all the biscuits in the lobby after all and we are quite late um so i think it remains for me to to thank everybody um our three presenters our six panelists um and you in the audience for your contributions and participation thank you very much i hope you found it worthwhile um thank you for attending and i'm going to just very briefly hand over to sarah and i with a quick housekeeping message thank you very much dale and thanks to you for having managed this session i just wanted to tell the people in this room that the next session in this room will be the service onboarding and catalog of services so if you were planning to attend this session you are in the right room if not you have the links link to the agenda or in your email from yesterday or here on the slide okay have a nice break so the next session starts in 15 minutes or maybe owin can think about have five extra minutes for the break leave it up to you owin we can start a few minutes later it's okay okay