Rittman Mead - Delivered Intelligence

Time to catch up on the second half of the week at
ODTUG Kaleidoscope. Tuesday was the one day that I wasn’t due to present, but due to my rather
over-enthusiastic volunteering I was still down to be an ODTUG Ambassador for
three presentations. This basically involved giving out the handouts and
evaluation forms when people came into the presentation, introducing the speaker
and then collecting the forms in at the end, not too gruelling a task but it
does mean you’re "on duty" all day without much of a break. In the end I "ambassadored"
twelve sessions, a bit of a record I think and a number that I’ll probably cut
down if I come again next year. Anyway, the first presentation I looked after
was Jean Pierre Djicks talk on "Why You Should Consider OWB for Data Modelling",
which was basically about using OWB to model your logical objects - dimensions,
measures etc - and the physical implementation - tables, partitions, indexes and
so on. The thinking here is that OWB 10gR2 supports far more of the physical
data warehouse features that are now present in Oracle Database 10g - though not
Index-Organized Tables, as a decision was taken not to include them due to the
performance benefit not being sufficient - and therefore it makes sense to do
all your modelling in OWB, rather than use it just for ETL and use a tool such
as Designer or Erwin to do the modelling instead.

I broadly agreed with the crux of JP’s argument, certainly you
can make the project a lot more "agile" if you use the same tool to model,
deploy and build your ETL routines; the only bit of feedback I gave was that
it’d help if OWB clearly separated the logical model (dimensions, measures,
hierarchies etc) from the physical model so that you can create a clear
separation between the two phases of development - also, Kent Graziano made the
good point that it’d also help if OWB came with a proper diagrammer similar to
the one in Designer, such that you could create a blank canvas and then include
logical or physical objects in the diagram as required.

Next up was the main presentation on Oracle BI Suite Enterprise
Edition. This was probably the most useful presentation of the week for me, as
it went into a lot more detail about the architecture of the Oracle BI Server,
how it processed queries and how it brings together data from heterogeneous
sources. Probably the most valuable thing though - and this is where the value
in attending events like this comes in - I spent a fair bit of time with some of
the guys who came over from Siebel, and in fact worked for nQuire before that -
and had a chance to discuss how the product works, where the "value" is and so
on. I also spend a bit of time on the following day in the hands-on labs talking
about the best way to put together the Enterprise Semantic Layer, and picked up
some good tips and techniques to take back to the office and pass on to our
customers.

The next presentation was Keith Laker’s talk on reporting
against the OWB repository. Keith, like the rest of the OWB team, is a pretty
smart guy and he came up with some interesting approaches and scenarios for
reporting against the public views over the Warehouse Builder repository. He
talked about two types of reporting - static "point in time" reporting that
provides information on the number of mappings within a project, how many
invalid objects are in the repository and so on - and more sophisticated
time-based reporting, that adds a time dimension to the information in the
repository and allows you predict, for example, how long your batch will take to
run at some point in time in the future. Keith went through some examples of
where Application Express and SQL Developer can be used to provide the static
reports, and a Discoverer Business Area that provided access to time-series
based calculations. Apparently the code examples and Discoverer Business Area
will go up hopefully on the

OWB Code Exchange area on OTN, and maybe also on the

BI SIG area of the ODTUG website.

I had originally agreed to do one more ambassador session later
on in the day, for Keith Laker’s second OWB presentation, but in the end I
decamped over to the hotel bar with a few other people and watched the England
game. So far I’ve watched the first game against Paraguay in Greece, the second
on against Trinidad in England, and now this third one in the United States, and
I must say it was pretty surreal to hear American commentators commenting on an
England game - lots of statistics on the screen (what does "0-1-1" mean next to
a team name?) and some very garish sets and excitable commentators throughout
the game. Of course in the end we drew, but went through, and I ended up staying
in the bar until the evening’s entertainment started - I’d been "on duty" since
Sunday so I decided to give myself a break for a few hours. The entertainment
sounded beforehand like a bit of a nightmare -
"Capitol Steps", an American musical
political satire troupe, which sounded to me of something like a cross between
Richard Stilgoe,
Pam Ayres
and Rory Bremner - actually turned out to be rather good; jokes about George
Bush, Donald Rumsfeld and Dick Cheney are fairly universal these days although
I’m sure six pints of Sam Adams
beforehand didn’t do any harm. Anyway, a pretty good afternoon and evening
although it did lead to a bit of a hangover the next day.

Wednesday started off with a fuzzy head, and what better to wake
me up than being the ambassador for Don Burleson’s talk on the Cost-Based
Optimizer. Don warned us at the start that he was also a licensed auctioneer and
therefore he could go on for an hour without stopping talking (very loudly), and
he certainly didn’t let us down. Unfortunately I had to bring my laptop in and
do some work on my demos for my presentation later on that day, but it was good
to catch up with him and Janet and certainly he had the biggest audience that
I’d seen all week. I also did the ambassador thing again later on that day for
Mike Ault’s ETL Techniques for Developers presentation, a very good presentation
and again a good-sized audience who gave him some pretty good scores on the
evaluation forms.

My final presentation of the week was at 1.00pm on the
Wednesday, on "Building an Effective Data Warehouse Architecture using
OWB10gR2". I had the usual problem with demos not working quite as they should
do just before the presentation, but when the time came they worked pretty well
OK and I think I gave a more relaxed performance than my XML Publisher one
earlier in the week. Thanks again to all those who came along, and if you’ve not
got the papers and slides already, you can download them from here:

• 
  Building an Effective Data Warehouse Architecture using OWB10gR2 (paper)

• 
  Building an Effective Data Warehouse Architecture using OWB10gR2 (presentation)

Going on to today now, I spend a few hours this morning catching
up with work things, and then got the metro down to the city centre to do the
tourist thing. I took the Red Line down to Union Station, then walked over to
the Capitol to start a slow walk through the National Mall to see the monuments.

Over about three hours, I went to the Old Post Office Tower
(thanks to Scott Spendolini for that tip) to take in the views, then walked down
to the Treasury, the White House and then over to the Washington Memorial.

Then, finally, after walking from the Washington Monument
through the World War II Memorial, up along the Reflecting Pool and along to the
Lincoln Memorial, I finally found a stall selling water round about five
minutes before I passed out with heat exhaustion - it was about 95ºF
in the afternoon and it’s very humid in Washington. Still, it was good to get
out of the hotel at last and you can’t come to Washington and not visit the
Mall.

That’s it for now though. I’ve got my BI Suite Enterprise
Edition training CD loaded up on my laptop ready for the flight back tomorrow,
the alarm set for 4.45am (ouch) and a flight to catch at 8.00am in the morning.
It’s goodbye to Washington, and hello to the Brighton to London train next
Monday morning.

Posted in User Groups & Conferences |