Hi All,
It is interesting to see the discussion around 'real-time data warehousing'...
One of the definitions (offered buy Inmon, Imhoff, Battas) is as
follows:
"The operational data store, or ODS, is a dynamic architectural construct designed specifically for high-speed, integrated operational processing. It achieves at the operational level what the data warehouse does at the strategic/managerial level."
So, a Information Architecture (someone doing ODS/DW and BI, we have too many acronyms) might consider that when the problem at hand is an operational issue, which must be addressed in the near term, and does not require a 'meeting of the minds' to discuss and set strategy to decide where large investments might be made, then some style of operational, integrated, information might be the better answer than try to make the DW 'fit' the operational needs.
Please remember, the reason the ODS developed was exactly because people started using the DW for operational processing and the technology of the day was unable to cope with the varying processing loads of OLTP and DW style queries. Further, as we evolved to archive/analytical layers and the increased use of dimensional models the processing requirements to get the data into a dimensional model further aggravated the cause of keeping everything 100% up to date all the time.
To the best of my knowledge, we are still at the stage where large organisations cannot reasonably expect to integrate operational processing with DW processing on the technology that exists today.
Plenty of vendors will tell you that the 'next release' will do this....IT people buy the 'next release' at their own peril.
Sure, we will see this distinction/separation between OLTP in the ODS and the DSS in the DW blur over the coming 5 years.....
But I am not aware of any absolutely compelling reasons to combine them today given that the data can be made to flow between ODS/DW relatively quickly without forcing it all into the one model...
If you look at examples provided for real-time decision making, they are minor and relatively un-important in the over-all schema of a large company. Most such decisions can be made just as well reporting driven from the operational system, or an ODS, or perhaps even with just a little common sense.
I recall a story recently (by a vendor) where the example was the 'real time' DW could assess that a customer had a poor interaction with the company and he/she could be called back and somehow placated.
And my response to that was that customer facing people should be trained to take reasonable care of customers so you don't need to spend millions of dollars catching the failures of the customer facing staff.....what a radical concept, good customer care..;-)
Just an opionion....;-)
Best Regards
Peter Nolan
Data Warehousing Consultant
Mobile: +353 879 581 732
Homepage:
-----Original Message-----
From: owner-dwlist@datawarehousing.com
[
Sent: 30 July 2004 19:36
To: dwlist@datawarehousing.com
Subject: Re: dwlist: Real time DW and the speed of real decisions
Please visit our sponsor today at
For help with list commands, send a message to <
From: "Greg Della-Croce" <SolutionBuilder2002@hotmail.com>
Gabriel,
Thanks for the input. The example you used is great. It falls
into
what I call the "Operational Information" for a client (as opposed to
"Management Information" or "Strategic Information"). The dashboard in
your example is taken from current activity, and may or may not have any need for data from last shift or yesterday or longer.
The short life span of the data's importance leads me to think more of
OLTP or ODS processing, not data warehouse. Is my view of DW to narrow
here?
----- Original Message -----
From: "Tanase, Gabriel (GEI, GEIH)" <Gabriel.Tanase@ge.com>
To: <dwlist@datawarehousing.com>
Sent: Friday, July 30, 2004 9:07 AM
Subject: RE: dwlist: Real time DW and the speed of real decisions
>
> provider of real-time data integration and resiliency solutions.
> Please visit our sponsor today at
> to access data warehousing white papers and best practices.
> For help with list commands, send a message
> to <
> word "help" in the body of the message.
> From: "Tanase, Gabriel (GEI, GEIH)" <Gabriel.Tanase@ge.com>
> I believe that it's not managers who are the supposed targets of
real-time
> DWs.
> AFAIK, it is telesales people who - allegedly - need analytical
customer
> profiles and contact/purchase history.
> If all sales people were so rational and analytical...
> I can, however, offer a genuine example for when managers/supervisors
need
> to take action within a half-hour from when a dial goes red,
definitely
well
> before end of day.
> In a high-volume call center environment (insurance, banking, customer
care
> etc.), if the ring-to-respond interval or the call length go
consistently
> "red" for more than half an hour or an hour on certain queues/teams,
it
> means that one has too few people allocated to those queues / teams,
or
that
> something systemic prevents people from closing calls in a reasonable
time
> e.g. the computer system they're using is too slow.
> The result is that too many customers are kept waiting for too long,
which
> is "baaad" (unless the company earns a share of the revenue from the
phone
> company based on call length).
> A call center supervisor / manager *must* take decisions within one
hour
to
> reallocate people or address other aspects of the situation. Otherwise
> her/his department will miss the SLAs with their internal or external
> customers for that day and there may be penalties.
> Now, whether this type of report must come from a DW is a separate
question.
> I guess that, in this particular case, it depends on the capability of
the
> call centre's phone switch software packages to provide real-time
reports
> at least "traffic lights". They usually do, so this kind of a report
doesn't
> really need to rely on a real-time data upload from the phone switch
> database into a DW, then for the DW portal to show the colors.
> However, if one has already built the real-time DW upload, then a
'traffic
> lights' report in the DW portal, refreshed every few minutes, should
be
> easy.
> Regards,
> Gabriel Tanase
> IT Systems Designer
> GE Financial Insurance Europe
> GEIS Shannon, Ireland
> e-mail: gabriel.tanase@ge.com
> Opinions expressed in this message are my own and do not represent the
> opinions or policies of GE or GEFI or any of its other employees,
> directors, officers, shareholders or affiliates.
> > -----Original Message-----
> > From: Greg Della-Croce [
> > Sent: Tuesday, July 27, 2004 10:30 PM
> > To: dwlist@datawarehousing.com
> > Subject: dwlist: Real time DW and the speed of real decisions
> >
> > From: "Greg Della-Croce" <SolutionBuilder2002@hotmail.com>
> > I may be a little dense, or maybe it is the type of clients I have.
> > Either way, can we discuss the >>Real<< value of real-time DW? I
am
> > obviously missing the value proposition here. I have found that
> > business people make very few decisions/actions on information on an
> > up-to-the-minute set of information. End-of-Day
> > information, yes. But
> > would a manager take an action on information based on the dash
board
> > gauge having moved from green to yellow in the last 10
> > minutes, OR would
> > they wait and see what the situation was at EOD?
> > Thanks!
> > Greg Della-Croce