[
Lists Home |
Date Index |
Thread Index
]
You
should ask Oracle about that. It will vary by business model.
For example, is the technology
a
shrinkwrap product for an established customer base that will easily and quickly
recognize it,
or is
a hard sell required? Is it custom software based on a requirement but
with the implementation
using
'innovative' features such as XML? Is it RFP-driven and what are the
terms of submission
(eg,
can options be substituted for features)? All of that.. but the easiest
thing to look at is the
release cycles for products. How often is a features
release made vs a bug-fixes release? Is
this
averaged across an industry where that industry is classified by releasing
common product
types? How long is it from the time of receipt of an RFP to the
BAFO/BARFO (Best and Final Offer/Best
and
Really Final Offer) to the contract signing? Then what is the
ARO (Time After Receipt of Order)
for
delivery (and this has to be scheduled with the Release Cycle). Release
cycles subsume
development, certification, packaging, etc., so it is a much longer cycle
than people suspect.
In
other words, most of the information you are looking for varies by business
model. Some companies
such
as Oracle have deep and very broad budgets for releases because they also have a
deep and
broad
customer base. However, Oracle delivery of a release is followed by the
cycle time as
described above of the middle tier vendors who build applications on top
of their products. And,
don't
forget that Oracle is only one part of the framework. Above the DB
sits the client/server
system
for the application. And if Oracle offers it but Microsoft SQL
Server doesn't, the feature
is
slowed for adoption because the RFPs sometimes do and sometimes don't let the
vendor
select
the DB system. For that reason, ODBC and middleware are used to map even
simple
things
like primitive data types. In other words, even if Oracle puts it in 10g,
it has to be
incorporated into the release cycle of the application vendor and this
party responds to the
content of the RFP.
I'd
like to say this happens in linear time, but only from a high level does that
happen. The
closer
you get to the actual code being written, the more elastic does the schedule
become
and
the more local. That is why the numbers I gave you seem
large. They are actually
somewhat conservative and based on having existing product to work
with.
I
don't object to RDF or any CG-style triple. It is that the more of
this we see, the longer
it
will be from the time it is a draft spec or standard to it being procurable, so
where systems
are
mission-critical (gotta go right now or sooner), the more likely it is that the
exotic
features will be ignored. Any specification that a customer
really wants has to be fit
into a
multi-dimensional scheduling view that is about as reliably predictable as a
wave
function operating under the exclusion principle. The more the Feds
push down
from
the top, the simpler that push must be because the States and the local agencies
can't
buy often or much. Federal grants are used to procure local systems
and
the money being used to keep forces overseas is coming
out of
budgets that would otherwise finance infrastructure. The budget
meltdown
combined with the deficits (and all of those forces) mean that
within the timeframe
I
predict we will see these technologies enter the near-time procurements, it is
likely
they
will be axed and simpler solutions will be tendered.
Remember Bosnia? It killed a lot of technical innovation, and it
was cost-trivial compared
to the
current adventure. So when I ask about roll-out schedules, I am
deadly serious.
I
suspect that before the middle of next year, there will be some hard and
detailed
questioning of HS-related initiatives, so the technological wishlist has
to be practical
and as
simple in implementation as possible. We have a long and bloody
thing to
do and
we have yet to really start securing homeland systems in the way they must
be for
the next stage of a conflict in which all we have yet seen are the opening
moves.
len
Len,
I am quite interested in statistics on
the timeframes from product availability
to implementations. Do you have any relevant data? For example, how long it
actually took from the time major vendors like Oracle started to
support XML to first
implementations?
I also
wonder how much (if at all) release timings matter. Some people I
talked to say that most government sites are still running on Oracle 9i, but
plan to move to 10g once Version 2 is available. The logic being that they
never go to version 1 of the major release. It just so happens that 10g
Version 2 is the upcoming release with RDF support. I guess we will start
to see soon what impact it
has.
As far is GJXML goes, it effectively invents its own
version of RDF with things like:
-
<xsd:complexType name="RelationshipType">
<xsd:attribute name="name" type="xsd:QName" use="required" />
<xsd:attribute name="subject" type="xsd:IDREF" use="required" />
<xsd:attribute name="object" type="xsd:IDREF" use="required" />
</xsd:complexType>
Regards,
Irene
Thanks for the information, Irene.
So if you are seeing the betas
now, procurement catches up in say three years,
implementations appear at the earliest, two years after
that.
At
five years ahead, the costs for this won't be seen until the first
half
of the first term of the next administration. And that is definitely
synonymous with Federal.
That
doesn't answer the questions of what we do with the rdf:id that
we wouldn't do
with
another semantic for an id. Let's talk for a minute about the
impact
of costing systems using these specs. Remember, it's useful
to have
an
RFP say "Comply with" without some means of showing how to comply
with
it.
GJXML is a good example. Today the RFP says "comply with"
and
then leaves it to the local procuring agency to figure out how to determine
compliance. The GJX... IEP is better because it
requires the actual subset work
to
be performed, a URI-identified instance of the schema to be produced
as a
Reference Schema, and possibly, the agency or State level instance
of
that. Now we have a definition (eg, incident-AL.xsd) that can be cited
and
validated. Note carefully: every one of these processes is
bid
at
consultant rates every time the customer asks for it. Until the
agency-
selected IEP emerges, the system procured remains in implementation
phase and does not cutover to live operations.
How
long will an agency wait before the system goes live? Answer:
about six to 12 months, but usually less than a year. Local
budgets
and
political priorities won't take the pressure of long
rollouts.
The
world is festering with meta-specs. They are decidedly expensive
items to chuck into an RFP.
len
[len] Show the
support for RDF in the major commercial software
frameworks.
Next version of Oracle 10g to be released this summer
has RDF/S built in. I can attest that they've done a good job at
supporting the standard (we've been beta testing the product). Some OWL
support is currently being worked on.
Software AG (vendor of XML database Tamino)
supports RDF and OWL. They've recently integrated Ontoprise's Ontobroker
engine into Tamino.
Adobe has had RDF support for quite some time
now.
I've heard rumors that IBM will announce something by
the end of the year - but these are just rumors.
Regards,
Irene
| 