Posted by Mike Gilpin on January 16, 2009
There’s been a lot of kerfuffle about Anne Thomas Manes’ blog post
declaring SOA dead.
The main thing that bothers me about declaring an architecture
“dead” is the idea that all the good actions that Forrester and other experts have
advocated people should be taking over the years – which are about the
practices of the architecture, not purchases of infrastructure – can be called
into question. What about all those years in which Anne Thomas Manes was,
herself, an ardent advocate of these same practices? Is she saying she was
No, she’s really just talking about the term no longer being
fashionable, about it no longer seeming appropriate to fund large
infrastructure investments under the banner of SOA. In her own blog post she
(weakly) makes the point that the practices of SOA should continue, without
spelling out the acronym.
This is just pure silliness. SOA (the architecture, not a
marketing banner for large infrastructure purchases) continues as before. For
years we’ve been saying there are larger and more strategic ways of looking at
architecture, mainly Digital Business Architecture. All of that is still true.
A more appropriate way to position the state of SOA is to say
that it’s become a well-accepted part of the landscape. Back in 2002 when Carl Zetie and
I published a report titled The
Post-Internet Era, we weren’t saying the Internet is over, any more than
post-Modernism is about Modernism being over. Rather, it means that thing has
become a part of the landscape, so we no longer really notice it. Now we look
back at the Modern design in furnishings in Mad Men (apologies to those who
don’t know this TV series) and the only thing that’s unique about them is their
concentration in one place to the exclusion of other things. Today these same
pieces would be at home in many of our homes, but would likely be mixed with
designs from other eras.
I’m not advocating the actual usage of the term
“post-SOA,” because we learned from using the term “post-Internet” that most
people don’t understand this usage, and thought we were saying the Internet was
dead. In fact, what we were labeling
was many things that are now labeled “Web 2.0,” although we missed the social
dimension (which is a pretty big miss! – although it hadn’t really become a
major factor when that report was written).
What I am suggesting is that SOA should just be
positioned as a useful school of thought, which produced some classics, but
which is not the last word in architecture, any more than object orientation
was. Hey, I’d love to have a Wassily chair in my home! But I like my other
The furniture metaphor only takes us so far, however, as it’s
not a good architectural approach to just mix a bunch of tastes randomly. It’s
more a design aesthetic that says we can mix the Wassily chair with the Breuer
chair with the Louis XIV side-table, if the rug and drapes have the right
colors and tie it together in a pleasing way. And the room has to be right, too
– and Alexander could tell us the classic patterns to which that room design
should adhere, if it is to be pleasing as part of the overall aesthetic.
Bottom line: SOA is no more dead than
objects are, or COBOL. SOA was (in Forrester’s view) never about buying a bunch
of technology. Your SOA Platform strategy is a set of choices for how you will
architect and design applications and infrastructure to enable them to work
together the way you want them to. Not an RFP.