EnterTheGrid/Primeur magazine:The most important recommendation of the report is to form a permanent "European Grid Agency". It should develop a Grid computing reference model, and reference implementation and take care of IPR issues. What is your opinion about such an agency?

Steven Newhouse:I am always very nervous about global encompassing organisations in the Grid space. Grids are a federation of services and resources. So the idea that one Grid agency in a sense can mandate or dictate or co-ordinate everything appears unlikely. Certainly there are issues that need to be tackled. But there is work going on in the Open Grid Forum and other bodies, and it makes sense for European industry and research to directly engage in those established bodies, rather than to set up a European specific agency.

Concerning the reference model: It would be very hard to get a single reference model that is going to be of use to all potential consumers. You probably want several composable models that can be easily implemented and incrementally developed - e.g. security, jobs, data movement, etc. This can be driven by industry and it can be much better described in terms of specifications and implementations than two or three years ago. An incremental approach can be user driven and has a greater chance to succeed than a large all encompassing reference model.

OMII-UK is working on software components that fit into this picture. OMII-Europe is working with its partners (Globus, Unicore, and partners in China) to look at interoperability aspects. It will look at assessing the interoperability between the different middleware services. A defined compatibility test allows other players to join the party with their own particular software implementation and the OMII-Europe repository will enable them to see how their software complies with the various specifications that the Grid community sees as being important.

EnterTheGrid/Primeur magazine:Although Europe did have the lead in Grids for academia and research, that does not seem to have resulted in a European Grid computing industry or a faster uptake in industry in general, the report concludes. Is that, in your opinion, a correct conclusion?

Steven Newhouse:Yes, I think that is a partially correct conclusion. There may be no specific European Grid computing industry but there are some examples of uptake in pharma and finance. The reason is that what academia sees as Grid computing is very different from what industry sees as Grid computing. They are almost two distinct use cases with only small areas of overlap. In academia the issue has always been about how to share resources between organisations. Within industry there is some activity in that area, but much more is focusing on maximizing the use of resources within their own organisation. That is a different problem. Some tools can be reused, but there are many different issues. Industry may use Grid techniques to explore dynamic service deployment or provision capacity oriented activities. This is another direction within the Grid space.

EnterTheGrid/Primeur magazine:The industrial focus in in-company Grid is that resulting from the way companies work, or is there a technical problem? Are intercompany Grids too difficult to do or is it they just do not need them?

Steven Newhouse:Both. Partly there is no clear need to drive this activity in industry except in a few collaborative projects and it is also perceived as having a much higher business and technical risk. The experience we have in building research Grids between different organisations is that this is very hard. Industry sees a clear need to rationalize internal resources and although that is not necessarily risk-free, the risk is much more manageable.

EnterTheGrid/Primeur magazine:According to the report the IT industry is very different from those in other parts of the world. In general, European IT companies are either large integrators, or SME's. Large European vendors are lacking. This requires a different approach to accelerate Grid computing uptake, hence the European Grid agency that could act as a focal point and handle IPR issues beyond the possibilities of SME's. What is your opinion about this?

Steven Newhouse:If there is no European Grid industry, I don't believe it is because there is no European Grid Agency. If there is a business case for doing it, industry will do it - and there is plenty of European participation in global company Grids. In my experience there is a lot of industry activity in the Grid where it makes business sense to do so. Companies like IBM will sell you a Grid solution if you request it.

You could argue that a company like IBM is not a European Grid IT company, but we live in a global world, so I am not quite sure whether the European branding of IT companies makes any sense.

Linking the perceived lack of a European Grid industry because of IPR issues is in my opinion wrong. If companies want a certain solution they will develop their own implementation based on existing standards or they will generate their own IPR to provide a solution that satisfies their customers needs.

http://www.hoise.com/primeur/07/articles/monthly/AE-PR-01-07-48.html</a> Do you agree with this? Or do you expect the European market to be much bigger or smaller in five years time?</em>

Steven Newhouse:Are Grids going to grow in the future? Yes! But to that amount - I don't know. Whether it will be 9 billion Euro's or something else I cannot predict. Related to the question as to whether Grids will grow is also the definition of what is a Grid. Grids in the future will be different from what we see today. There will be much more emphasis on the dynamic provision of resources and services, and getting 'ring fenced' access to resources behind the service. Some companies are already prototyping services along these lines. The current EGEE is a combination of national resources where the Grids are provisioned manually, in the future that will be done much more automatically.

EnterTheGrid/Primeur magazine:If one reads the report, one sees that a big part is devoted to IPR and licensing. To get Grids that cross enterprise boundaries, or even to have large heterogeneous Grids within a company, IPR and licensing of components have to be clear. The report opts for a style of Open Source Software license, but says this should be investigated. It also says the legal situation in Europe is different from, for instance, the USA. For instance in software patents and waiving liability. What is your opinion about this issue?

Steven Newhouse:There are really two issues here in my opinion: The IPR of the software that constitutes the Grid middleware and then the licensing issue around the application software that people will use as part of the Grid. Concerning the first part: I do believe the IPR of the actual middleware itself is a not a significant issue in the sense that people will buy software solutions and those software solutions will start to interoperate with each other through standards, because this is what customers are demanding.

Licensing is becoming more of an issue as the Grid community grows into new areas. In the UK we find people who want to use (say) the Matlab environment as part of their Grid jobs. They need to know the environment at the remote site so their jobs will run.

So not only is there an issue about automatically provisioning that Matlab environment remotely, but then because Matlab is licensed software you need to be able to use and run that software on the remote machine, but ideally using your own existing licenses from your home system. This is posing a lot of problems within the UK infrastructure at the moment.

So the companies need to look at their license policies and how they can be managed through a software infrastructure. There appears to be a growing need for a more flexible Grid-friendly licensing model; whether it will be pay on demand for a specific time period on a specific machine, or pay per invocation or something like that. The traditional license model where it is tied to your machine, your desktop or your IP address is not really viable in the Grid space as application execution become much more mobile.

EnterTheGrid/Primeur magazine:What are your experiences with OMII-UK thus far? Although it is primarily aimed at eScience, perhaps something can be said about providing commercial grade software too. It seems like OMII-Europe is also focused on part of what a European Grid agency is intended to do? What in your opinion is the most important obstacle to Grid commercialisation in Europe (if there is any)?

Steven Newhouse:Within OMII-UK our experiences to date have been very good. It is clear that people want software that is easy to use, easy to install, easy to maintain and all of these issues. These are clear requirements in the academic space: people do not have the time to battle through installing software as this is not their "business goal". This software is supposed to enable them to do science.

Our role in OMII-UK is to raise the quality of the software provided and used in the academic space - towards the quality of commercial grade software. We aspire to commercial grade software, but we realise that we don't have the resources to do that, so what we are doing is to make sure that the software that academics are using is better than normal academic software.

OMII-Europe is more about interoperability and certification of software both in terms of its quality, such as robustness and portability, but also as to the interoperability of the software. Within OMII-Europe we are also using the outcomes of another European project called ETICS to provide a testing infrastructure.

Our strategy is to describe what the software does. For instance, when you buy a car, you get a checklist where it says: the headlights passed a test, the tyres another test, etc. It is up to the customer to decide whether they buy the car or not. In the case of OMII-Europe the tests and test results are collected and viewed through the repository.

One of the most important obstacles to Grid commercialisation is with having well defined and widely adopted standards. The reason I believe why industry has not engaged fully is because it is not clear what the sense of the market is, or the way to get access to the market. Standards are a very good way of indicating what the market is and its relative importance to the community.

A major challenge of Grid computing is the assembly of standard components from various domains and presenting them to end-users as a coherent whole.

EnterTheGrid/Primeur magazine:Any concluding general remark?

Steven Newhouse:It is clear that the expectations around Grids have to be managed. They have already been shown to demonstrate new ways of working to some communities. The challenge is to now expand these benefits of Grid into new communities. So, Grids certainly have a future, but their delivery date for general users is still several years out. The efforts that are going on to support researchers have real potential and have already shown some benefits to global collaborative use of e-infrastructure.