Thursday, September 24, 2009

Roving reporter: Grid perspectives in four questions

As part of this historic week – the last and final EGEE conference – we've been asking people for their view on the past and future of grids. We've posed four questions:

1. What do you think EGEE's main achievement has been?

2. What were its main challenges?
3. Did grids develop they way you expected?

4. If you had a magic wand, where would grids be in 5-10 years?

Here is some of what we've heard so far . . . .

= = = = =

Vangelis Floros, EGEE

Main achievement? I think bringing all these people and countries together to build this infrastructure, uniting technologies, and bring scientific communities together – this is really great. You can see the success from these conferences, the people that you see year after year, the collaboration – this is really the essence of grids.

Primary challenge? Bridging different demands, trying to exploit technologies which were in infancy when we started and trying to satisfy many scientific and political requirements.

Did grids grow as expected? No, absolutely not. It was very positive to see grids growing so much in number, but the actual efficiency of them is an issue that still remains. And some of the architectural and technology choices that were made at the beginning, maybe they were not the best ones. But this is it, so I think grids are slightly different than they were expected to be in 2000.

Grids in the future? Well, I'm not sure we are going to be using the same name. I think that we will use more cloud-type technology, as it shapes up and matures. I think that the whole landscape of computing will be combined, loosely-coupled clusters integrated with high performance resources, and we'll see something much bigger and more unified.
= = = = =

Erwin Laure, KTH

I think the main achievement is the community building. EGEE was able to get an international community together, not only of resource providers, but also users. That's the main achievement – it's not really technical, it's sociological.

The main challenge? The sociology! Working amongst all these people. If you try to develop a technology across the whole of Europe, or even world wide, you face many strong voices that want to push it in one direction or the other. Finding a compromise among those voices is a challenge and we are not always able to succeed in that.

Did grids develop the way I expected? Yes and no – we have this global infrastructure and it is being used, so in that way they developed as I expected. However, looking back, there is no fundamental difference between the way things are run today and the way they were run when EGEE started. The scale changed, but fundamentally it is still the same thing. I think we lost a few opportunities in moving grids – or let's call it e-Infrastructure, which is a better term these days – in to more a more sustainable, easier to use, easier to manage mode. So in that way things didn't develop in the way I was expecting them to.

Where will grids be in 5-10 years? That's a difficult one. Looking back, if things continue as they have for the last 5-10 years, then I think we would still be as we are now, and that probably cannot be. I think that grids will move from this tightly integrated, huge infrastructure that we have at the moment – on many kinds of levels – to a more flexible, independent component, where applications will pick and choose, and grids will be a very thin layer. At least, that is how I hope it will be.
= = = = =

Manuel Delfino, PIC

I think EGEE has demonstrated that you can take these basic technologies that were theoretical six years ago and you can deploy them to do something which goes well beyond Web or basic internet technology.

The biggest challenges were probably the fact that there was so much pent up demand for doing this. People had the internet, they could imagine what they wanted to do with it, but they didn't have the tools. So we had to deal with many user requests, and very high expectations, for something that could only be very basic at the beginning.

Did grids grow as expected? In some sense yes, in some sense no. I think using Virtual Organizations, and the way we group users, developed more or less as I was expecting. But we still have to deal with being able to quickly create small VOs.

We still have some technical problems to solve, which have to do with being able to bring jobs to where the data is, because otherwise it is very inefficient. That, and what I mentioned before, being able to quickly create VOs – and to apply this grid technology to very basic things. As an example, being able to create a blog which is seen only by one VO.

Please share your own thoughts below!

No comments: