Posted by Theodora A. Varvarigou 24 April 2009

Virtual World interoperability is currently a hot topic in the research world. There are at least two major initiatives trying to tackle the issue by setting up a standard for enabling interoperability: the joint effort from Linden Labs and IBM within the auspices of the Internet Engineering Task Force (IETF), and the MPEG-V, which is taking place within the auspices of the International Standardisation Organisation (ISO) with Philips driving the effort. At the same time there are numerous virtual world platforms based on different implementations (and some widely used open source ones such as Project Wonderland and OpenSim) and a great number of users which is constantly increasing. This is only an indication of the virtual world dynamics and in turn, how enormous are the technical challenges, especially for an application that is being traditionally based on well-established software engineering technologies such as multi-tier (client-server) architectures.

But let’s talk a little bit about these challenges. Apart from the presentation issues that arise when someone attempts to port an avatar from a virtual world platform to another which are more related to 3D modelling and 3D-engines in general, there are other more hardcore challenges such as data, identity and license schemes management, security, privacy, trust and common interfaces issues. And addressing these aspects is crucial if one wants to achieve end-to-end interoperability coupled with viable business models.

For one, data aggregation is required for dealing with the variable data formats and structures that each virtual world is using while secure data transmission is necessary to protect sensitive private and commercial information. Then, it is the common API for application development in several virtual world platforms. Moreover, even though actual user identity is not always reflected in virtual worlds -mainly because of current vague registration and authorization schemes-, the identity that is attached to the avatar must be carried throughout the virtual island hoping adventure. Finally, a trust framework must be established between virtual world providers which are often antagonistic, by allowing them to gain control over the type of information that they are giving out as well as the other quality terms that govern these relationships.

However, a more careful study, will lead the meticulous reader to the conclusion that all the above seem to be aspects of the more general problem of complex system heterogeneity, scalability and business relationship management. This, coupled with the fact that virtual worlds are by nature service-based applications, rings many bells to the researchers that work in distributed computing. Several of the abovementioned problems have been addressed to a large extend by Grid computing, i.e. data aggregation, end-to-end security, trust establishment and SLA management, functional interoperability through common infrastructure services and the list goes on. BEinGRID alone covers many of these issues. One has only to check some of the technical solutions in Gridipedia to find out that we can start building an interoperability middleware for virtual world interoperability using these as a baseline SOI.

The bottom line is that problems which have been addressed by Grids (even with functionalities exposed through Clouds) seem to remain all-the-rage and re-using such solutions even partially makes great sense. It is not necessary that it will worth the effort, but it is necessary to investigate it.

Posted by Mark Parsons,  14 April 2009

Grid standards were supposed to make it simple to create interoperable service-oriented Grids. However, very few global Grid standards have been agreed and even fewer have been widely adopted. Why is this and what does it tell us about the global IT industry today?

The early part of this decade saw an explosion of Grid research projects trying to understand how to build a new generation of distributed computing applications which would revolutionise science and business. I was involved in this process from the beginning – helping to write CERN’s DataGRID proposal in 2000 and writing and leading a series of UK and European Grid research projects and spending many millions of Euros in the process.

From the beginning the funding bodies insisted we devote a considerable amount of our time to “standards setting”. Ever keen to ensure our proposals were funded we dutifully undertook to focus on standardisation as a key exploitation outcome. I now wonder if this was the right choice.

Many of the people I’ve worked with over the past decade have spent countless hours pursuing standards in face-to-face meetings around the world or teleconferences that have either never come to fruition or, when they have done, have been so diluted as to be pointless.

It was clear that early implementations of Grid middleware were poorly written lash-ups; written to prove an idea rather than deliver product quality software. By proposing the Open Grid Services Infrastructure (OGSI) in 2003, the Global Grid Forum (largely led in this case by the Globus project and IBM) tried to move from a non-standardised distributed computing framework to a web service based platform which could be extended over time. This of course upset everyone. I now believe that this was because people wanted to experiment more before they were constrained into the OGSI standards.

It’s very interesting to look at how HTML came about. People had been playing with hypertext for many years before Tim Berners-Lee used the ideas to create HTML 1.0. At the outset HTML wasn’t a standard but it was rapidly picked up and used by so many people that it became one by default. Given the economic value that has ensued from HTML, I don’t believe it could be created today by a standards process. None of our large IT companies would sit down and allow any of their competitors to gain such advantage and the discussions would be endless.

In many ways this is what has happened to Grid standards since OGSI. Many in the IT industry saw OGSI as an IBM-inspired land-grab. All of the major vendors then marched their troops into the standards arena – largely sidelining the researchers who had conceived the underlying ideas in the first place – and halted all appreciable progress in Grid services standards thereafter. WS-RF and WS-DM tried to bring everyone together, but Microsoft and Sun countered with WS-Management. A grand plan was hatched to bring WS-DM and WS-Management together but this has still to see the light of day.

So what does this tell us about the Grid? It tells us that the distributed computing community hit upon an interesting research area at the turn of the century. So interesting in fact that the major IT vendors felt sufficiently threatened to spend large amounts of their money debating standards that now seem pointless. The world has moved on. Those of us who deliver real-world Grid computing solutions use what’s available – and we learn more about how to use these pieces of the Grid jigsaw puzzle every day. We also know that standards are important but only after the technologies on which they are based have been proven – not before – and certainly not because one IT vendor or another says so.