Utility computing - a myth or emerging reality?
Harold Ainsworth
07/08/2008

The term utility computing evokes the idea that all sorts of computing resources are a plug-and-play, pay-as-you use commodity just like electricity. In utility computing, customers do not own any hardware (such as servers or storage systems), software (enterprise applications, for instance) or other kinds of IT resources. Instead, they “rented” these from a technology service provider and are charged accordingly based on their usage levels.

The potential benefits of utility computing (also known as adaptive or on-demand computing) are certainly attractive from both the customer and vendor standpoints.

On one hand, customers neither have to acquire new hardware for meeting drastic short-term surges in their computing requirements, nor to dispose of the new purchases once the demand subsides. This offers cost savings, reduced footprints and proper technology-to-business alignment and other advantages.

On the other hand, vendors can potentially net a wider clientele by offering them flexible, customised usage packages based on a common set of hardware, software and services. Furthermore, they only have one centralised pool of IT resources to administer, maintain and upgrade, instead of having to handle one or more IT infrastructures per customer.

Sounds like a win-win situation, right? Well, it’s time for a reality check.

Utility computing in its purest form is still some way off, although it is edging closer towards fruition. There are still a number of obstacles that the industry needs to overcome.

For utility computing to become commonplace, customers need to accept that IT resources have to be centralised into a common offsite pool that is managed and operated by the service vendor. This inevitably means that customers will lose some control over their use of these resources – which they no longer own, but merely rent. That is utility computing in its simplest form, but in reality some customers still ask for facilities – such as data centres – to be housed onsite for their exclusive use. The privacy or security concerns fuelling such requests are understandable, but customers with dedicated setups cannot realise cost savings through economies of scale.

The main driver behind utility computing is demand volatility – sudden, short-term spikes that are followed by rapid plunges back to, or near, the original levels. For some organisations, demand volatility may occur regularly and is therefore easier to accommodate. However, demand volatility can sometimes be due to external or unexpected events beyond anybody’s control, thus making it hard to manage. For example, many airlines experienced drastic plunges in their passenger loads following the September 11 terrorist attacks in 2001 and the SARS crisis in 2006. The smaller loads naturally led to a drop in their IT utilisation needs – but were they able to quickly adjust their usage levels accordingly?

I will not answer on the airlines’ behalf, but consider this – according to Gartner, 70 to 80 per cent of IT costs are fixed while only 20 to 30 per cent are variable. High fixed costs restrict organisational flexibility because users are generally unable to quickly modify their utilisation levels following business downturns – or upturns – caused by external events.

Fujitsu has rolled out initiatives to help customers lower their fixed IT costs. For example, we have delivered the following results for a mid-sized European airline customer:
- shaved 30 per cent of overall costs
- reduced the proportion of fixed costs to overall costs
- introduced round-the-clock service availability
- conceptualised plans to further raise service levels

To achieve this, vendors and customers must rethink the conventional – and often adversarial – approach to outsourcing characterised by the “fixed price mechanism”, rigid Service Level Agreements (SLAs) and other hallmarks. New relationship models are needed so that both parties can resolve contractual issues in a win-win manner, especially since complex problems typically require extensive knowledge-sharing before they can be solved.

Gartner views utility computing as having five distinct maturity levels, each with different telltale signs:

Level 5 (most mature level) – characterised by business automation (i.e. increased business agility)

Level 4 – characterised by policy-based automation and provisioning (i.e. improved IT service delivery)

Level 3 – characterised by virtualisation and portability (i.e. optimised hardware and software deployment as well as increased efficiency)

Level 2 – characterised by automation, rationalisation and resiliency (i.e. reduced costs, improved service levels as well as enhanced resilience)

Level 1 (least mature level) – characterised by centralisation and standardisation (ie. economies of scale)

Their assessment is that most organisations are still hovering around level 2. This illustrates my earlier point that utility computing in its purest form is still some way off.

Based on Fujitsu’s experience in Singapore, customers are slowly becoming more receptive to services that will allow them to climb the maturity ladder – but it is still very early days.

- Harold Ainsworth is Practice Lead, IT Governance & Value Management, Fujitsu Asia.

advertisement