Innovation UK > News > Innovation UK Vol5-1 > Service oriented computing - a new wave of innovation

The journey that the Grid Computing Now! KTN has undertaken in the past four years has moved on now to a new stage, where organisations are broadly familiar with the opportunities that large scale distributed computing can offer, in terms of scalability, capacity and resource efficiency. The key question is: ?To what end can we improve the deployment of our computing services to meet the escalating expectations of our customers/partners/ suppliers??

The emergence of the major internet service providers, Google, Amazon, e-Bay, Apple and many others, has changed our expectations of the services we expect to be offered in several profound ways.

• We expect all services, whether private or public sector, to be accessible online. From purchasing a new book, music or video; through the purchase of travel tickets; access to repeats of broadcast TV programmes missed; access to essential government services; renewal of road tax; submission of tax return, etc.
• We expect services to be accessible online at any time, day or night, year round. Amazon?s peak service period is in the immediate run up to Christmas, e-Bay and Apple see peak periods emerging immediately following where people seek to exchange presents.
• We expect services to be fulfilled in a reliable and timely fashion. These huge peaks of demand are to be met without disruption, meaning not only online capacity to meet demand, but also the logistics infrastructure to ensure that presents are delivered on time and at the last minute. Otherwise, the website is not visited again! Our industry leaders have been fostering this reputation as an highest priority.
• We expect services to be accessible from anywhere, office, home, mobile phone, internet café, in a secure and appropriate fashion. Yet we expect our online presence to be appropriately secure, without the risk of privacy or identity being compromised!

Ian Osborne

So who do we mean by ?We?? ?We? is no longer the privileged few thousand computer specialists, but a huge swathe of society familiar through e-mail and web browsing with the use of the internet and the commercial services available online. ?We? can be numbered as being between 5% and 50% of the population. That is between 3 million and 30 million attempting to access the services on offer. ?We? can be characterised as people from a very young age to substantial old age, wrestling with the computer interface, looking for communication, information, entertainment, services and enrichment. ?We? is not normal in the traditional term, but a skewed distribution with a heavy weighting towards the younger end of the spectrum.

The younger half of the population definitely view the use of services as normal and our society is adapting to this with ticket collection machines at concert halls, theatres and cinemas, for online purchasers. Our airlines are depending upon web booking and printing of boarding passes at home. Our train companies are providing us comprehensive online booking facilities with options for ticket delivery. ?We? no longer book our holidays in the high street, but directly with companies offering the widest of choices of destination, routes and facilities. ?We? now look for new cars online, searching through the car manufacturer website, identifying preferred choices, testing them at car dealerships and then either buy them online or use the online price to extract a better deal at the showroom.

This revolution in our purchasing and information seeking behaviours has been facilitated by the new wave of service providers, who have understood these rules and built their systems to meet the needs of a customer base spanning to 100s of millions, worldwide. A very different model from that of the traditional ICT shop, where the systems have been sized for a few hundred, or thousand users, and have happily existed online, and off, for the traditional working week.

Typically these systems have occasional peak periods of operation and this has also been taken into account during sizing. Thus lying dormant for most of the time ? and hence a relatively low utilisation figure of between 10% and 15% for server equipment is achieved. It?s actually worse on the desktop, where PCs and laptops can be considered to be basically idle, even if they are being used at the time!

So what?s to be done? First stop for many people has been to improve their utilisation and we have seen virtualisation in the computer room make great strides in this area in the past three years. At GCN! one of our first case studies, Auto Trader, pointed at an improvement towards 60% average utilisation. Such an approach has become widespread now as organisations adopt virtualisation tools with or without grid capabilities to achieve this ?quick win? and reduce their capital spend and energy bill. Some companies have extended this to the desktop too. Although there is a more delicate balance to be struck between driving desktop equipment harder, increasing heat generated in the office, versus relocating processing to the data centre in a more efficiently managed environment. See GCN!?s Merrill Lynch case study. And, of course, mobile computing equipment may sometimes be moved!

These measures help drive up efficiency, but say nothing about the scalability of the applications in use. A parallel innovation we have tracked at GCN! for the past four years is Service Oriented Architecture and the service enablement of IT applications in general. Early thoughts seemed to point to individual organisations reshaping their application portfolio through ?exposing? services from their existing applications. These services could then be accessed by web browsers and combined to form new applications, possibly enhanced though the inclusion of newly adapted services. Another early GCN! case study, Cattles, pointed the way to this through the use of industry standard tools.

If you want to understand the rich potential of this style of computing, then look no further than the excellent example of the online retailer Amazon. With Amazon, your home page is completed with the fulfilment of between 100 and 150 service requests, dynamically completed as you watch. It is in this way that you receive pricing updates on the items as yet waiting for purchase in your shopping basket. It should be remembered that this information is being provided for you at the same time as millions of other users online. And that the underlying infrastructures span the globe ? incurring the penalty of transmission delays ? without you really being aware of them. The e-Bay application is similarly complex, with hundreds of millions of accounts, a totally diverse range of products on offer ? some are auctioned for consumption on a sameday basis, again with a global infrastructure. You might ask the questions next time you sell, how on earth did that item appear in the product catalogue so quickly? I wonder where it is stored?

How are these applications able to scale these peaks of demand and activity? Well they are service enabled and the infrastructures that deliver them are tailored to be resource efficient, highly resilient (a short failure may well incur many thousands of pounds lost profit), and highly adaptive to the demands of the customer. Well, you?re thinking, that?s ok for them, but we don?t have the resources, time or money to make this change. Nor do we have the need! But is that the case? Mid last year e-Bay reported a customer base of 268 million accounts. This is an European Commission-sized customer base. An opportunity for government services to be more efficiently delivered? Amazon?s peak demand period in the run up to Christmas drives their entire strategy for development each year. Is that really substantially different to the flurry of activity which takes place when tickets for a major rock concert go on sale; or the last-minute demands for online submission of tax returns?

The key point is that while we are being more successful in harnessing our computing applications to provide a more efficient compute environment, a similar and more substantial change must take place on the application side if we are to capitalise. Particularly with the excitement that Cloud Computing and the idea of on-demand computing is generating across the industry. There is little point in creating efficient infrastructures if we cannot see how best to exploit their capabilities.

Service Oriented Computing provides the key. The major application vendors have been diligently working away at delivering integrated service platforms which enable a more dynamic delivery of information throughout the enterprise. Leading vendors in the space have been snapped up to provide the integration toolkits required. There are applicable standards too. In the wider world of the web, services are already being made available, including the ubiquitous Google Maps, and other information services of that ilk, which simply invite the combination with other information such as location-based data (your supplier network); pricing information (eg online estate agents); traffic routing information; etc.

Your customers, users and collaborators are all beginning to use these tools and services more. How will you adapt your business applications to make them better able to scale and integrate with their needs? Grid Computing Now! KTN will be highlighting this area of interest in its agenda looking forwards. Keep in touch with the GCN! community, review the published case studies and background materials. We are shaping our future agenda around the concept of scalability to help you achieve your business objectives in this key area!

Further information about the
Grid Computing Now! KTN and
its activities can be found at:
www.gridcomputingnow.org