| 3.1 No Quick Hit Despite all the potential benefits, some of our clients have raised the
contrary question of whether or not XML makes sense at all. This is a refreshing question,
for answering it requires analysis from first principles, free of the hype currently
surrounding XML. The quick answer is that in the short term, it probably doesn't make any
sense. XML offers no dramatic gains in efficiency in any one area that one can point to it
and say, "see, there's the value!" Rather, the benefits of XML tend to be
diffuse and cumulative rather than concentrated and specific.
We compare XML to SQL databases in several respects. When the
first relational databases came into general release, similar questions were raised as to
whether the benefits justified the cost. The benefits of the relational data model and of
SQL data access are also diffuse and cumulative, and therefore hard to quantify while the
costs are obvious. Standard data files store data just as well as relational tables, and
provide better performance in many cases. SQL also provides a loosely coupled data
interface into application programs in a way that is very similar to XML.
Indeed, we have identified no technical goals that require
XML for accomplishment. Just as VSAM stores data just as well as DB2 or Oracle, and C++
produces applications that operate as well or better than those written in Java, so fixed
record formats can do everything that can be done with XML. The point is that the
cumulative gains from the many small efficiencies of a relational database over standard
files proved its value after large-scale adoption, and the same will be true of XML.
What is required for a realistic, balanced appraisal is a
focus on the cumulative gains over time, not the single "quick hit". If the
standard is to be a quick payback, then your XML initiative, arguably, should be focused
only on meeting external requirements and perhaps on new application development, with
little attention paid to retrofitting XML selectively into existing applications.
On the other hand, if the business standard to be applied to
XML is to be optimized value creation over time, (both through cost reductions and through
enablement of new business opportunities) then a broadened perspective emerges. In this
case, we would recommend a strategy of balancing theoretical research and development with
practical demonstration projects that are not throwaways, to provide the practical
experience that a thoughtful manager wants without a wholesale commitment. Neither theory
nor practice should be too far in front of the other, given the rapidly evolving nature of
the XML standards and the need for disseminating practical knowledge of XML throughout the
IT organization.
In essence, this approach combines a leap of faith that the
cumulative value will appear in time to provide a positive return on investment from the
efforts, with sufficient near term reality checks to allow mid-course corrections or
outright abandonment if warranted. In our opinion, adopting an "XML ready"
application package does not really address this issue, although it does expose members of
the staff to XML technology. We recommend that small projects involving locally maintained
programs be defined, perhaps involving only single programs, to explore this technology
and evaluate how to ensure a positive return on investment over the medium and long term.
Pragmatism and sensible cost/benefit analysis are the order of the day. |
