Let us focus for the moment on one particular aspect, which I believe will have the greatest impact, the lowly world of 'cycle scavenging'. Across datacenters where servers are sized for peak performance but stand largely idle and on desktops (which by nature are rarely used continuously let alone at full capacity) we have precious IT assets with very low yield.
If we were to consider the sunk costs and environmental impact alone of this wasted capacity we would have a compelling argument to change this state of affairs.
Yet if we were to go further I am sure we would be avalanched if we asked the business what new calculations would you perform, what new pieces of business intelligence would you drill into, what new aspects of customer behaviour would you investigate if you were given almost free computing power at your fingertips.
By the very nature of modern IT infrastructure, the majority of unused devices run Windows in one form or another. This realisation prompted the creation of the Avanade Grid Asset (AGA) to safely and simply distribute tasks, by taking full advantage of native Windows management features and resource co-ordination through Active Directory groups.
Organisations such as Monte dei Paschi di Siena (MPS), one of the oldest banks in the world, have seized on such grid technology initially as a more affordable alternative to extending mainframe capacity for traditional jobs, and latterly to open up a stream of new innovative applications including critical regulatory reporting for IAS and Basel II (see the bank's recent paper presented at the 1st International Workshop on Grid Technology for Financial Modelling and Simulation: Proceedings of Science Grid 2006) .
To fully usher in the Grid-frame era, we should remove the blinkers created by previous HPC scenarios, which account for a very small percentage of business needs and are largely addressed by dedicated clusters. Instead we should begin the dialogue with our end users who want ubiquitous business intelligence and insight, what Gartner terms 'sense and predict' from realtime infrastructure, and soak up the already purchased compute power under our noses.
Most importantly - grid computing and the value it brings is for the masses - not for PhD programmers working with abstract toolkits on staggeringly expensive hardware.
These ideas will be discussed on Day One of the Architect Insight Conference, with an overview of the MPS case study and thoughts on the evolution of Windows grids with Compute Cluster and Excel Services.
Richard Hall, CTO, Avanade