I think that negative reaction to 'integration' in the SOA sphere is because of
high risk of getting "two cities become connected via roadways" in the way
nobody could join the roadways in between and one can get into the roadways
only being in one of the cities (following Todd's example). Though it not
necessary manner of integration, the integration concept does not preclude from
such individual solution.
In SOA, the integration task (yes, it does exist) is formulated based on reuse
of integrating channel and, correspondingly, the city entrances. It is assumed
that there may be a third city which might be connected to the first two ones
via the same roadways. It is done by design, not by requirements change; this
is the major difference. Mediation is just one of the techniques for join the
roadways (round-about or rotary). I do not think it is about functional vs.
non-functional aspects...
- Michael
Todd Biske <[EMAIL PROTECTED]> wrote: I'll
bite. I don't consider integration a dirty word. Where I think
it has bad connotations is where integration is done by throwing a
whole bunch of "glue" in the middle. Integration will always be
necessary, although the conceptual model we desire is one of
connection, rather than integration. When two systems need to talk,
we connect them and off we go. That's certainly an over-
simplification, as the information domain will never be so
standardized in the way that roads, railway, power, etc. are to allow
it to be that simple, but we continue to try to minimize the
integration effort to just the parts that are absolutely necessary.
Your point about mediation versus integration is also an interesting
one. Mediation definitely doesn't have the same connotations as
integration, although it's certainly a grey area as to where
mediation ends and integration begins. It's almost a question of
where non-functional aspects end and the functional aspects begin.
If we go back to the role of connections, two cities become connected
via roadways. The roadways are mediated by traffic signals, etc.
Some type of mediation will always be necessary as part of our
connectivity.
Just my $.02.
-tb
On Jul 26, 2007, at 5:59 PM, Rob Eamon wrote:
> Many articles, blogs, discussion forums, etc. seem to
> treat "integration" as a detestable creature, to be avoided if at all
> possible.
>
> I suppose one's view of integration, positive or negative, depends on
> its definition. I imagine some will state that "integration"
> encourages data duplication and reinforces application silos. Others
> will pose that assembling services isn't "integration" but instead is
> some other less offensive thing, like composition or such. Or that
> services from different ownership domains are connected using
> mediation or gateways or such, not (horror of horrors) "integration."
>
> I'm quite interested in hearing views on "integration"--that it's
> good, bad, indifferent, meaningless, or otherwise.
>
> My POV is that integration is a broad term, encompassing many, many
> styles of interaction between entities. If two applications are
> communicating with each other, they are integrated (with no comment
> as to the quality of that integration). If services are
> communicating, especially through an intermediary of some sort, they
> are integrated and the tools and mechanisms used to enable that
> interaction are "integration tools."
>
> -Rob
>
>
>
>
> Yahoo! Groups Links
>
>
> [EMAIL PROTECTED]
>
---------------------------------
Get the Yahoo! toolbar and be alerted to new email wherever you're surfing.
