Hot Topic: Making IT disappear

In his keynote presentation at InfoWorld's CTO Forum in San Francisco in April, Irving Wladawsky-Berger, IBM's VP of Technology and Strategy, defined an objective for the IT industry. He characterized the 'end' of the technology uptake cycle by pointing out that "…people [will] login to a virtual service…the goal is to make it…boring…". In other words, technologists will know they've succeeded when the fruits of their labours are so simple, reliable, available and predictable that we all just take them for granted.

This vision of IT as another utility has been a developing theme in the industry for a number of years. It is interesting today because the industry is in the process of progressing from the easy, over-the-horizon futurology versions of this vision, to plausible roadmaps that may soon be influencing mainstream IT infrastructure investment decisions. IBM's eLiza, Sun's N1, HP's Utility Data Center, Compaq's Adaptive Infrastructure, and Microsoft's .NET are all examples of products, projects and thinking that are contributing to this progress. In the case of HP, they've even built the utility concept into the product name!

In the last five years, Quality of Service (QoS) has been high on the agenda as a key factor in achieving 'utility' status for IT. Assume, for a moment, that adequate standards will exist and communication infrastructure will be able to deliver appropriate bandwidth. Even then, it is not until QoS can be controlled and guaranteed that an IT service can be used, and ignored, in the way we use other utilities. After all, we'd probably all buy our own generators if we couldn't predict the voltage, frequency and power-availability of our domestic electricity supply.

A lot of work has been done to manage QoS for communication links. But of course this is only one part of the jigsaw puzzle. From a users point of view, it is QoS right through to the application level that is key to the value of the 'service'.

It turns out that to manage application level QoS efficiently, it is necessary to build 'instrumentation' into application code. Without this instrumentation, the operating system must treat the application like a black box, and use coarser controls - for example, dynamic allocation of resources - to achieve target QoS levels. The language used to discuss the required instrumentation includes terms like 'sensors', 'actuators', and 'probes' reminiscent of the physical world of machines and electronics. Happily there has been some industry-wide discussion of the nature of this software, ranging from IBM and HP's joint proposal for an 'Application Response Measurement' API in 1996, to the Open Group's creation of the QoS Taskforce in February 2001.

All this technology is moving out of the backroom laboratory, and becoming credible as a factor in investment decision-making.

For example, with eLiza, IBM is rolling out a range of system management capabilities, some new, some translated from tried-and-tested mainframe technologies. As part of eLiza, IBM has announced their plans to publish to ISVs what they must do to take advantage of the capability to deliver defined QoS in this way. And this isn't just some esoteric frontier of technical development. IBM has also initiated market awareness activities - a full page colour advert in the Wall Street Journal in February with a headline 'You don't have to buy it to use it' was evidence of this interest in a 'utility' approach to delivery of IT capacity.

ISVs must think about how to respond to this new dimension in the interface to the operating system. File system, user interface, communication, and perhaps licence-control and system management interfaces are familiar, but, for many ISVs, QoS instrumentation is new territory.

My advice is seize the opportunity with both hands. At Cambashi's 2001 seminar, we went against fashion and noted the remarkably low proportion of global software revenue that would flow through the Application Service Provider channel during the following three years. At the same time, we pointed out that there would come a time when third-party ASP-like remote delivery and the pay-as-you-go model would dominate enterprise software revenue streams. Our forecast required us to rearrange the digits of the year (2001) to say when this would be! 2010 still looks a good forecast. So if ISVs think about QoS instrumentation in 2002, and build prototypes in 2003, there will be time in 2004 to think through the major business, sales, and marketing issues. By 2005, there will be two groups of corporate users who are ready to talk and, from 2006, become the 'early majority' of users of external, IT utility service for their core applications. These two groups are firstly the users who have been organizing their IT as an 'in-house' service (whether outsourced or not), and secondly the users who have been using remote services to handle peak demands.

ISVs who have followed the timetable above will be able to send out sales reps who can approach these groups with a clear understanding of what they can offer (i.e. application-level QoS guarantees). If this turns out to be a differentiator from the ISVs competitors, it could be enough to open the door to competitive accounts. Why? Because the costs will be so much lower - all the arguments rehearsed by ASPs over the last few years, largely based on the core of economies of scale, will be presented in a package acceptable to the IT mainstream.

This will put IT well on the way to achieving the disappearing act forecast by Irving Wladawsky-Berger. Always-on infrastructure will be delivering what people want to pay for, i.e. always-available, predictable, reliable applications. And we'll be able to ignore it!

Peter Thorne
email: peter.thorne@cambashi.com

Also in this issue:

Feature Article: Cambashi Seminar Review
John Dwyer, recently voted business and professional columnist of the year 2002 by the Periodical Publishers' Association, has written a synopsis of the presentations given at the Cambashi Seminar held at the Gaydon Motor Centre in April…

Book Review: Exploring Online Magazines
Webworks:e-zines Exploring Online Magazines, Martha Gill, Rockport Publishers,
Mike Evans reviews this book and ends up disappointed in Cyberspace…

Cambashi researches best practice and assists IT suppliers in best practice implementation. For more information on Cambashi services please email info@cambashi.com

e-Xpertise in Industry operates as an opt-in email. The opt-in policy means we never intentionally send any email to anyone who does not wish to receive it.

To subscribe: send an email with the word "subscribe" in the subject line to : expertise@cambashi.com

© Copyright 2003 Cambashi Ltd

back to top