So long, silos, hello!
- By Jack Vaughan
If the mythical Rip Van Winkle were to awake from his 20-year snooze not at
the end of the American Colonial era, but rather at the end of the Internet
Gold Rush -- and yes, that is a big if -- he actually wouldn't find all that
much changed. Software marketers these days zigzag the country with presentation
foils strangely similar to the ones they carried in the mid-1990s, when enterprise
application integration was briefly the buzz.
The problem is that the departments in our corporations are not truly connected
in a great web of productive endeavor. Connecting data and applying logic to
an app just one PC away is not much easier than it was before the first Arpanet
machine messaged another machine. When a software firm has to show a big end
market to its venture capitalist backers, the problem can appear as a multimillion
In the days of office automation, when Xerox Parc was developing its great
ideas that have now been commercialized, the phenomena might have been called
"islands of automation."
Over the years, the terms "stovepipes" (also used to describe Digital
Equipment's general approach to innovation) and "silos" were
used to describe this essential problem of disconnectedness. If your eyes were
to glaze over as software marketing execs described the problem today, they
might surprise themselves and again recall the "silo" metaphor of
long ago as they outlined the problem.
Inmon and Zachman are prominent, but they are just two of the many industry
experts who have described the silo problem ... which still needs solving. What
is odd is that it serves as something of a Rorschach test. The problem reveals
much about the minds of the makers of software, but its solution remains hidden.
The problem of integrating silo systems is approached differently by makers
of OO software, application integration software, BPM solutions, CRM software,
content management software and so on. That so many Web services vendors are
now singing its merits as an integration technology (rather than as the next
killer Web app), adds fuel to the fire.
It does not help matters that Web services vendors have a hard time looking
at what they do without seeing it as the last stage in civilization's evolution.
When pressed, they admit their loosely coupled methods may not suit everybody,
everywhere right away.
But silos are everywhere. Of late, a stream of content management software
makers have wobbled out bloodied but unbowed by the Internet bubble burst after-blast.
And silos are their targets because their approach is decidedly document-centric,
whereas much of the activity in the distributed computing realm has focused
If there is ever to be a breakthrough here, it might require the technology
world to answer a simple question: "Is a bank check a transaction or a
The question can lead to hot debates these days at beer blasts where people
-- well, some people anyway -- argue the merits of the Semantic Web put forth
by Tim Berners-Lee and others as the next step.
Silos seemed to go away, and have now reappeared -- leading a Skeptical Examiner
to say: "So long, silos, hello!"
* * *
The difficulties of measuring software -- of describing what it is, for that
matter -- have always made it a hard trade to ply. And since the days of the
first electronic ledger, there has never been a shortage of obfuscators -- those
who would tout their product beyond all reason, or denigrate competitors.
This can often be done without actually lying; with technology, it is usually
a question of emphasis. The opposing technology evangelists simply apply "spin"
to actual facts.
Microsoft's recent launch of Visual Studio .NET provided the occasion for a
storm of spin. Few vendors in tools or architectures could avoid the temptation
to "make news." Their target was usually the .NET architecture.
Another recent spin inflection point: BEA's efforts to present its vision of
Web services at its yearly user conference.
This constant spin -- which often mangles the truth -- is an enemy to us all.
The Skeptical Examiner asks for a cease fire.
What's your take on things? Is a bank check a transaction or a document? Has
the storm of spin gone too far? Let me know.
Jack Vaughan is former Editor-at-Large at Application Development Trends magazine.