Work by IBM could bring on-demand computing power to the
business
In last week's feature I mentioned that IBM's Web services
resources seem to be aimed almost entirely at developers. This is
not to denigrate the quality of these offerings. As the list of
developer features with a
Web service angle indicates, IBM's activity here is
nothing if not thoroughgoing.
Among the more technical papers, there are also some that will be
suitable for a general readership, notably a good series of
articles called
Web services insider. Another useful document, if
slightly more formal, is a description of IBM's
Web Services Conceptual Architecture.
Alongside all these words, there is plenty of code, too. The best
place to keep up to date with IBM's work in this area is
www.alphaworks.ibm.com/webservices.
The free
Web
services toolkit, with more details
available.
IBM is also active in proposing new components for Web services,
notably in the form of XML applications. For example, there is the
Web Services Inspection Language (click here for
background and
full specification) and the charmingly named
Web Services Experience Language, which is designed to
allow providers of interactive Web services to exploit multiple
distribution channels.
All this Web services activity bears witness to IBM's diligence,
but it is not something that will affect the average business user
of the Internet. However, a
recent
announcement about the company's collaboration with the
Globus project may well do.
As I wrote six months ago, Globus is fast becoming the de facto
standard for the new area of grid computing (joining disparate
processing centres with fast links to create a powerful virtual
facility). IBM has been stepping up its involvement with grid
computing as part of its general conversion to the virtues of open
standards - the Globus Toolkit is open source, and GNU/Linux
clusters are frequently employed as a cheap source of computing
power for the various nodes of the grid.
IBM's announcement possesses a double significance. First, it
signals an important convergence between grid computing and Web
services. The fit is good: both are about utilising distributed
resources in a co-ordinated and seamless way. In addition, both
must support heterogeneous environments, and depend on open
standards to provide the glue.
Moreover, both move closer to regarding the entire Internet as a
computing platform on which applications can be run - exactly where
and how is a technical detail that end-users do not, and should
not, need to worry about.
All these issues are discussed in an excellent document called
The
Physiology of the Grid. The title refers to an earlier
paper,
The
Anatomy of the Grid, written by essentially the same
authors, who are all top grid researchers.
The other significant element of IBM's announcement is that it
marks a step towards the "gridification" of its
existing product range.
Hitherto, grids have been chiefly applied in the scientific
computing domain, where software can be written specially to draw
on their power. IBM's latest move opens up the possibility of one
day running standard enterprise applications on grids. This would
provide business with true computing power on demand, supplied by
the digital equivalents of today's analogue utility
companies.
Next week: The ADSL era?