John Schmidt, from Informatica, penned an
interesting post titled: IT Doesn't Matter - Integration Does.
Yes, integration is hard, but I do profoundly believe that
what's been happening on the Web over the last 10 or so years also
applies to the Enterprise, and by this I absolutely do not mean
"Enterprise 2.0" since "2.0" and productive agility do not compute
in my realm of discourse.
large collections of RSS feeds, Wikiwords, Shared
Bookmarks, Discussion Forums etc.. when disconnected at the data
level (i.e. hosted in pages with no access to the "data behind")
simply offer information deluge and inertia (there are only so many
hours for processing opaque information sources in a given
day).
Enterprises fundamentally need to process information
efficiently as part of a perpetual assessment of their relative
competitive Strengths, Weaknesses, Opportunities, and Threats
(SWOT), in existing and/or future markets.
Historically, IT acquisitions have run counter intuitively to the
aforementioned quest for "Ability" due to the predominance of "rip
and replace" approach technology acquisition that repeatedly
creates and perpetuates information silos across Application,
Database, Operating System, Development Environment boundaries. The
sequence of events typically occurs as follows:
- applications are acquired on a problem by problem basis
- back-end application databases are discovered once ad-hoc
information views are sought by information workers
- back-end database disparity across applications is discovered
once holistic views are sought by knowledge workers (typically domain experts).
In the early to mid 90's (pre ubiquitous Web), operating system,
programming language, operating system, and development framework
independence inside the enterprise was technically achievable via
ODBC (due to it's platform independence). That said, DBMS specific
ODBC channels alone couldn't address the
holistic requirements associated with Conceptual Views of disparate
data sources, hence the need for Data Access Virtualization via
Virtual Database Engine technology.
Just as is the case on the Web today, with the emergence of the
"Linked Data" meme, enterprises now have a powerful mechanism for
exploiting the Data Integration benefits associated with generating
Data Objects from disparate data sources, endowed with HTTP based
IDs (URIs).
Conceptualizing access to data exposed Databases APIs, SOA based
Web Services (SOAP style Web Services), Web 2.0 APIs (REST style
Web Services), XML Views of SQL Data
(SQLX), pure XML etc.. is problem area
addressed by RDF aware middleware (RDFizers e.g Virtuoso Sponger).
Here are examples of what SQL Rows exposed as RDF Data Objects
(identified using HTTP based URIs) would look like outside or
behind a corporate firewall:
What's Good for the Web Goose (Personal Data Space URIs) is good for the
Enterprise Gander (Enterprise Data Space URIs).
Related