With Grid only just entering the Enterprise space and still plenty of developments required including for industry standards, the forward-looking experts of the IT industry are attempting to define the logical next step beyond Grid and use models such as Utility Computing. With the holidays’ period approaching, I could not resist the idea to give you a new acronym to play with in 2007: SOKU – Service Oriented Knowledge Utility. OK, fair enough: I already hear the comments saying that’s not brand new; that’s true. The idea of SOKU emerged around 2003 / 2004 as part of a work activity driven by a group of experts named the Next Generation Grid (NGG) in collaboration with the 6th Framework Program of the European Commission.

So what’s in SOKU? Let’s take a look at the context first. Grid, SOI (Service-Oriented Infrastructure), SOA (Service-Oriented Architecture) will gradually allow the provision of computing, data and information capabilities through specific managed services. Hopefully access to these services will be standardized across all components. Experts have identified the need in the future to close the gap between end-users (i.e. users of IT services) and Grid technologies with knowledge capabilities using semantic technologies. If Grid could be thought in multiple shapes and scales (e.g. computing and/or data, Enterprise and/or Collaborative), you should think big with SOKU, some kind of ubiquitous Grid approach with an evolutionary system made of very large numbers of interconnected nodes of any type across multiple business or society boundaries. From a software perspective SOKU is meant to provide a totally customizable environment, leveraging virtualization capabilities across the IT stack combined knowledge-assisted services to provide the automation and advanced functionality required. In a true service-orientation, SOKU architectures comprise dynamically assembled services where behaviour and location of software as well as data are adjusted on the fly. The utility aspects of SOKU refer to a predictable model where services can be called with specific functionality, performance and cost expectations.
The convergence of Grid, Web Services and Semantic Web is perceived as a fundamental step towards the development of commonly available SOI to create and provide access to intelligent services anywhere, anytime and in a secure, cost effective and trustworthy way. As Grid exceeds the computing & HPC solution requirements to move into composing intelligent services allowing ad-hoc dynamic creation of IT organizations for businesses or else, future Grid and SOKU developments have clear and important implications for security, scheduling, data movement, applications/executables movements, messaging, transactions, right managements, licensing… This new Grid operating environment would require some new levels of intelligence, notably based on semantically-rich based services, in order to provide the security, trust, performance and self-* properties required. As new applications are breaking away from monolithic executables running on large centralised platforms, they tend to become dynamic groups of computed entities. Such applications are modular and leverage heterogeneous sources of data, computing, network and software services. SOA’s have conceptualised this approach. The next generation of application will also rely on a set of infrastructure services to avoid having to re-develop the whole complex network of services required each time a new application needs to be deployed. This services combination model is similar to what we can find on the Web. So if the first generation of Grid was very much purpose focused (Compute, Data…) the next generation(s) of Grid will tend to become multi-disciplinary, collaborative and even more dynamic and distributed. This leads to an organizational model for IT in the Enterprise where business entities could be supported by a group of virtual organizations that will be dynamically populated by components, services and data (information).

So what’s the difference between SOKU and a next generation Grid? In addition to looking at the next generation Grids as the accomplishment of many of the original targets specified for Grid, next generation Grid are intended to allow a more seamless integration and interaction of multiple Grids, very much like in a Web approach. SOKU’s may be part of next generation grids and may also represent a collection of SOKU’s. The big difference comes from the representation of services. SOKU services are semantically described; i.e. the services metadata are built in a machine-readable format to ease their automated use. This allows the dynamic composition and configuration of services providing the autonomic behaviour required to achieve self-management. Furthermore, semantic tools can be used to describe the content of services; .i.e. services can process information based on reference representations of the appropriate type (schema, vocabulary, ontology). A SOKU approach does not impose that all services need to be semantically rich, but it requires that SOKU services are implemented in the SOI to handle information and/or knowledge processing. This can be particularly important when considering ILM requirements in terms of persistence, distribution, privacy and security in such dynamic environment.

The impact on data storage? It really hard to say for now as the industry is trying to tackle the existing challenges such as location transparency, security, data movement…to better manage distributed data repositories across and between Enterprise infrastructures. Addressing such elements would help defining the role of Enterprise Storage in the Enterprise Grid. But to achieve the level of abstraction and contextualization described for the next generation Grid and SOKU, the industry will need to think about how to store data in the future so that many of the management operations can be automated along with the mechanisation of services composition. Without the ability to automate many of the operations and therefore the ability to scale – a typical example is the discovery of resources and services– visions like the next generation Grid and SOKU won’t become reality. Peer-to-peer technologies may be of help here, in support of the self-* orientation of IT. No need to say that such vision would not exist if the management and communication entry points of such concepts are not standardised.

Is it or is it not a new IT era? The SOKU concept seem seems to point at the convergence and integration of multi-disciplinary IT domains such as Grid, Web Services, Semantic Web, Data Storage, Peer-to-peer, Lifecycle Management, Trust and Security…which have been studied for many years. The answer will certainly come from the research programs addressing these domains – in the meantime we can only admire the ambition of these concepts to lay stronger and stronger IT foundations to help driving business and societal projects.

Trackback URI | Comments RSS