BI Blogs

I’ve just been having a play around with the
Internet Archive Wayback Machine,
which holds "copies of the internet" going all the way back to 1996. Using the
site, it’s possible to have a look at how sites looked several years ago, and so
I thought I’d take a trip down memory lane and
see how my site
looked over the years - you can click on the links to go to a fully-working
version of the site, which is pretty cool.

Going in chronological order, the first screenshot on the left
is from

February 2002, the first serious website I put together where I mostly
focused on the work I was doing with Linux and Oracle. You can see from the
style of the page that I was in "Unix hacker" phase at that time, to be honest I
was more likely to be playing around with Linux in my spare time rather than
Oracle. On the right is a smartened up version of the site from
June 2003,
with much the same content but also, if you look closely, my first ever blog
posting, done through blogger.com but using my own template.

Shortly after starting out using blogger, I quickly found it was
a bit limiting (no categories, no RSS feed at that time) and moved it all over
to Radio Userland (the left-hand screenshot) in

September 2003. I quite liked the default template that it came with - this
was when blogging was really taking off and Userland / Dave Winer was at the
centre of the blogger phenomenon. In the end though I quite liked the
extensibility of Moveable Type and over Christmas and
January
2004, I reworked it all and move it to MoveableType 2.6, which was around
the same time that Brian Duff moved over from Radio Userland as well and started
up Orablogs, which still runs on MoveableType 2.6 to this day.

As 2004 went on I wanted to add a bit more content to the site, and put some
links on to articles that I’d written, and ones written by other people that I
found useful. In
July
2004 I added some HTML DB-style tabs to the page with this additional
content, and the site pretty much stayed that way until the start of 2006, when
I gave it a bit of a makeover, expanded the "resources" section to include some
write-up and commentary on Oracle’s BI toolset, added a forum and expanded the
set of links and book recommendations. The final screenshot shows how the sites
looks in

March 2006, which is probably how it’ll stay for a while now as it’s
increasingly getting a big job to change things around.

While you’re there, you might be interested how

Oracle.com looked back in 1996 ("Visit the new Network Computer web site
for the most recent NC news!"),
Slashdot
in 1997 ("Should Netscape GPL Mozilla - Yes or No?") and

AskTom in 2001 ("Sorry, I currently have a large backlog of questions.
Please try back later" - not much changed there then)

I worked for a client the other week that asked me to come up with some tips
and best practices for Oracle Warehouse Builder. Without giving the game away
(otherwise you’d never pay to hire me) here’s ten tips for working with Oracle
Warehouse Builder 10g.

1. Don’t skimp on the design and analysis phase. Just because Oracle
   Warehouse Builder is a graphical tool used for designing data structures and
   mappings, it doesn’t mean that you can skip the boring bit at the start
   where you do all the analysis and design. You still need to understand your
   source data, and you still need to specify how the data maps and how you
   deal with the transformations and cleansing. This is especially important
   when you consider that some customers use OWB as an opportunity to put the
   data migration work into the hands of less experienced developers, who won’t
   neccessarily be aware of design best practices and who won’t be aware of all
   the nuances of the source data. Every project I’ve ever worked on has said
   that they wished they’d been able to spend even more time on analysis and
   design, so don’t think that OWB means you can miss this stage out.
    
2. One the same subject, for each mapping produce a specification and test
   plan on to the developer so that they know, item by item, what bit of data
   maps on to where, and how to test whether their mapping is working ok. It
   doesn’t matter where you do this, it can be on a piece of paper, in a Word
   document, in a spreadsheet or in a powerpoint slide.
    
3. Make sure from day one that developer machines are sufficiently
   powerful. You need at least 1GB of RAM and a 1GHz CPU, and ideally your
   Design Repository database will be on a server with locally attached disks
   within your department, not on a SAN that’s being shared with your
   E-Business Suite implementation. Oracle Warehouse Builder is effectively an
   OLTP application that stores each change to the warehouse model as a
   transaction against the repository tables, and this application like any
   other OLTP application suffers when the disks it’s using are suffering
   contention.
    
4. Use the Flashback technologies in Oracle 9i and 10g to implement mapping
   and process flow "transactions", so that your mapping or process flow stores
   the SCN (System Change Number) at the start of the mapping, and then rolls
   either the tables within the mapping, or the whole database, back using
   FLASHBACK TABLE or FLASHBACK DATABASE if the mapping or process flow fails.
   What we do is use FLASHBACK table to roll back the tables within a process
   flow if it fails, and use FLASHBACK DATABASE if we want to roll back the
   entire ETL process. It’s certainly quicker than performing a point-in-time
   recovery if the whole ETL process goes belly-up and it gives us the ability
   to pull together a number of mappings and processes into a single atomic
   package of work which we can reverse out if need be.
    
5. Use collections to give yourself the ability to create daily ETL
   releases. Each day, collect together into a collection all of those mappings
   and other objects that have changed and been "checked out" by the
   developers, and use this to export those objects into the "Daily Build"
   environment…
    
6. Create a "Daily Build" environment which contains the ETL process as at
   the end of the previous day, and which includes all the changed items in the
   collection you created yesterday. Then, using an automated deployment
   process built using OMB*Plus scripts, deploy all of the objects and mappings
   and run a "smoke test"
   to check that the build hasn’t been broken. If it has, get it fixed, and if
   it hasn’t, you know that you can create an ETL release if you need to.
   Thanks to Jon Mead and Donna Kelly for the last two tips.
    
7. Record and review the run times of your various mappings and process
   flows. Once a week, list these out in order of run time, longest at the top,
   and in addition identify those whose run times have varied the most over
   time. Use the ALL_RT_AUDIT_EXECUTIONS and ALL_RT_AUDIT_MAP_RUNS runtime
   repository views to obtain your run times, and bring in additional
   statistics from statspack and the Unix server running the database to add to
   your diagnostic data.
    
8. When tuning a mapping, look for the simple answers first. Assuming that
   you’ve designed your mapping properly in the first place, the reason it’s
   running slowly is probably because an index isn’t getting used, or you need
   to increase the size of HASH_AREA_SIZE, or because you’re joining too many
   tables together. Only when you’ve tried the obvious try techniques such as
   
   tracing, the danger with trying the complicated first is that you get
   drowned in diagnostic data and miss the obvious solution.
    
9. Keep mappings simple, do one thing at a time. The danger with loading
   multiple tables, or loading one table after another after another, is that
   you’ll never be able to work out what’s gone wrong if the test figures come
   out wrong, and you’ll have the devil of a job tuning the mapping if you’ve
   got multiple levels of INSERT INTO .. SELECT in your mapping. Keep It Simple
   is the watchword here.
    
10. If you’re starting out on a big project, you’ve got lots of data to
    integrate and aggressive deadlines, hire someone who’s
    done it before to give
    you a hand before it all gets out of control. OWB only goes so far when it
    comes to a data warehouse project, and the real skill comes in apply
    software development best practices to what you’re doing, and knowing when
    to use Warehouse Builder and when to step outside of the product and build
    your own framework. Better to spend a few days upfront than learn all the
    mistakes the hard way.

Thanks to Jon and Donna for providing input and ideas along the way.

I have just received the following announcement on the final release of Oracle Data Miner 10gR2:

Oracle Data Miner is a graphical user interface for Oracle Data Mining that helps data analysts mine their Oracle data to find valuable hidden information, patterns,…

[ This is a content summary only. Visit my website for full links, other content, and more! ]

Thanks to everyone for taking the time to respond with comments and feedback (link) to my post on my request for features people would like to see in the next release of Discoverer (link to post). Some of the common requests were for automated scheduling & distribution, save-to-file, images in titles, more advanced conditional formatting, etc…

The good news is that the features people asked for are the ones that we have been hearing from customers, the sales force, and others - so the good thing is that there are no surprises there! The even better fact is that the new Oracle BI suite Enterprise Edition has all these features and a lot more. For example, you can do very nifty alerting and distribution using ‘Delivers’ in the Enterprise Edition, with the ability to send alerts to different devices.

There is a flip side to all this however - as you would have read in various web sites and through announcements (link to Google News) and on this blog itself (link to post, link to post on Oracle’s BI briefing) that there are now three editions of Oracle Business Intelligence: Oracle BI Suite SE One, Oracle BI Suite SE, and Oracle BI Suite EE. While Oracle Business Intelligence Standard Edition (link to page on OTN) comprises Oracle Discoverer, BI Beans, Reports, and other products, the Standard Edition One and Enterprise Edition suites are built upon the analytics server that Oracle got from its Siebel acquisition. These two suites offer a lot of additional functionality, including many features that customers have been asking for in Discoverer.

Separately, Oracle has also committed to making Discoverer essentially a data source to the analytics server, which means that for those customers interested in adopting the Enterprise Edition, they would be able to leverage their existing investments in the current Oracle BI stack. Those customers that want to continue using Discoverer will be able to do that also, as we have committed to providing enhancements on top of Discoverer, in addition to Oracle’s policy of providing lifetime support (link to the Discoverer statement of direction on OTN for more specific details). Also see Oracle’s Lifetime Support Policy FAQ document (link to doc).

Therefore, a question that assumes more importance is that from among the basket of requirements that customers have, which are the ones that they would specifically want to see implemented in Discoverer.

I have created a small poll - please take a minute (will take even less than a minute actually) to fill out your responses…
If I have left out any choices that you think are important enough, do drop me a line at abhinav.oracle at gmail.com.

this is not a commitment to deliver any functionality in any stated timeframes. Your responses shall help, certainly, and are an invaluable input into our planning. This blog, I hope, provides a quick way to elicit informed opinions from our customers, partners, and others interested in Oracle BI.

I had posted a couple of weeks back that Michael Armstrong-Smith had started a Discoverer blog (link to post, link to Michael’s blog). The blog is now fully functional, and with a nice look and feel (LAF in Oracle-speak), Michael has been busy adding lots of very useful posts. Also included are posts related to his latest book, the Oracle Discoverer 10g Handbook, with clarifications (link) and errata (link).

I know I had said I would be posting excerpts from the book, but between a viral infection, a holiday, wife’s trouble with a root canal infection and a wisdom tooth, daughter’s indigestion, a convocation, and other stuff at work, I haven’t had that much time. Hopefully things shall change for the better next month (which is tomorrow!!).

Did I say convocation? Yes!! I am now an MBA from the Indian Institute of Management, Bangalore (popularly abbreviated as IIMB) (see more on this blog). And a gold medallist at that (never hurts to do some self-publicity!).

People may have missed this: there is a new sub-site on OTN for “ORACLE FUSION MIDDLEWARE and Microsoft Interoperability” (link to OTN page), which brings together under one sub-site all of Oracle’s products and technologies as they relate to Microsoft Office.

There is also a new doc that has been released as part of the Oracle Application Server 10g (10.1.3) release (Part No B25781-01), titled “Oracle Application Server Developer’s Guide for Microsoft Office Interoperability” (link to HTML doc, link to PDF doc).

To prepare this document a virtual team comprising of product managers and development contacts from different Oracle product groups was assembled and each product team then contributed their time and resources to bring to completion chapters for their specific product (or technology). One of the takeaways that people should get from reading this document is that even today there are a lot of areas where Oracle and Office integrate and there are lots of technologies that can be leveraged to bring more of Oracle’s technologies and products closer to the end user using Office as an interface.

Here is a list of the chapters from the developer’s guide:

• Understanding Microsoft Office 2003 Extensibility Technologies
  Which covers such topics as XML Schemas, Smart Technologies (Smart Documents, Smart Tags, Smart Clients, Smart This and Smart That), Task Panes, Research & Reference Services, and more.
• Understanding Oracle Application Server Interoperability with Microsoft Office
• Part II then gets into the specifics of building Microsoft Office Interoperability technologies using Oracle products and technologies.
• Creating Smart Documents That Interact with Self-Service Business Processes
• Completing Forms and Entering Data Using Microsoft Office
• Securing Smart Documents and Web Services
• Delivering Business Activity Monitoring Alerts and Reports to Microsoft Outlook
• Delivering Business Intelligence Information to Microsoft Excel
  This is the chapter that covers Discoverer and all the rich and different ways you can get static as well as refreshable and live information into Excel - via plain CSV (Comma Separated Values), or Excel HTML, binary Excel with formatting preserved, or export Discoverer crosstabs to Excel as Pivot Tables, or the Web Query (iqy) format, or using the OracleBI Spreadsheet Add-In that lets you use the familiar BI Query Builder and Calculation Builder to build and edit queries inside Excel.
  
• Managing Tasks and Collaborating in Microsoft Outlook
• Provisioning User Identity Information and Alerting Microsoft Outlook Contacts
• Accessing in-Context Web Information and Invoking an Enterprise Portal
• Saving Microsoft Office Documents to the OracleAS Portal Content Repository
• Delivering Enterprise Reports to Microsoft Office with Oracle Reports

I was just doing a bit of random web surfing when I came across two
interesting articles on the state of the tech publishing market.

"The
Eventual Death of Developer Magazines" by Eric Sink looks at how developer
magazines are becoming increasingly redundant now that we get all our technical
information from the internet. According to Eric,

"Two days ago I stopped in at my local Borders bookstore and noticed
that not one of these magazines ["Software Development", "Visual Studio
Magazine"] was available for sale. They carry hundreds of periodicals.
They have an entire shelf of magazines for scrapbookers. They have niche
titles like Biblical Archaeology Review. But I couldn’t find a single pub
which was focused at software developers. What’s up with that?

The unavoidable truth is that these magazines have largely ceased to be
relevant. More and more, software developers get their information on the
Web, not from a magazine. As just one example, compare the quality of the
technical content in any developer magazine against
Raymond Chen’s blog. It’s
not even close, and Chen’s blog is pure content, as opposed to "tidbits of
content squeezed in between the ads".

It’s certainly the case that I don’t buy any developer magazines any more,
with the vast majority of my technical updates coming from sites such as OTN,
technical blogs and "free" magazines such as the OTDUG journal, SELECT, Oracle
Scene or Oracle Magazine. Of course Eric mostly works in the Microsoft
eco-system which has a different sort of developer profile than the Oracle
eco-system (probably a lot more newbies, a lot more people looking to implement
the latest new thing from Redmond) but certainly he’s right about this sort of
publishing increasingly becoming irrelevant. I don’t know how it works in the
States, but over here we still get magazines such as Computing, Computer Weekly
and IT Week but increasingly they’re devoid of recruitment ads, down to a
fraction of their late 1990’s size, and pretty much "old news" when they finally
get through to you.

On a similar subject David Heinemeier Hannson thinks that
the technical publishing
market is now redundant and moreover gives book authors a poor deal:

"I’ve been talking to a lot of friends and acquaintances who are
writing tech books for a wide variety of old-school publishers and I can’t
believe the deals they’re taking.

It seems that the industry standard is something akin to 10% of the profits
(which easily take 4-5-6 months to arrive), being forced to write in Word,
and finally a production cycle that’s at least a good 3 months from final
book to delivery. That’s horrible!

And what do you get in return? Usually not all that much. There’s rarely a
big marketing push to be had and you’re expected to do lots of the editing
yourself. So you get some editing, a cover/layout, and the distribution done
for you. Is that worth 90% of the profits and the torture of writing a book
in Word and then bouncing versioned documents back and forth?

The standard sugar coating of this setup is that you should not expect to
make money writing a tech book. That it’s not about the money, but the fame
and authority and satisfaction of seeing your name in print. While all of
those things certainly do have value, why on earth would you want to accept
the premise that writing a book is not going to be worth it for the money?!

Of course it’s not going to be for the money when you only land 1/10th of
the crumbs that trickle back to the publishers from Amazon and independent
retailers. Especially in tech land where its very rare to make it up on
volume (your book is not going to be in every airport around the world).

No, we have to change the game such that each copy becomes much more
profitable. Just accept the fact that you’re not going to sell 100,000
copies and optimize for the scenario that you might sell 5,000 or 10,000
copies. It is entirely possible to make money at those levels, but not if
you have a traditional cost structure."

David’s argument is that if the technical publishers adopted more agile
approaches to publishing, and reduced their overheads, they could afford to
offer book authors a better deal than they currently get. Moreover, if they
don’t start offering good deals, then it’s entirely possible for authors to
"self-publish" in the form of downloadable PDFs, and make money on the typical
5,000 - 10,000 copies a moderately successful book sells.

Tim O’Reilly came
back though with an interesting reply on why the deals are as the are, and why
self-publishing is only going to be viable for a small percentage of authors:

"At bottom, I believe that you’re reasoning from insufficient data,
David, and generalizing from experiences that are not necessarily
reproducible by other authors. While established publishers are often
sclerotic and slow, they aren’t all stupid or venal. As businesses get
larger, they have to manage their economics by the averages, not by the
successes. If we could all pick winners every time, we’d all be rich and
living in Lake Wobegon …

.. First off, to clarify, royalties for most tech books are typically
10-15% of net sales, not of "profits." That’s a big confusion. Most book
publishers have profit margins in the 5-15% range, and so 10-15% royalties
paid to authors are actually equal to a 50/50 split of profits …

… How about sales volumes? David’s got some false assumptions there
too. Very few computer books sell 100,000 copies. Most traditional computer
book publishers are in fact already optimized to be profitable on books that
sell from 5-10,000 copies. That’s the size of the market for most books when
the publisher is able to call on all the traditional avenues of
distribution. And even then, the odds aren’t that good. We get a report from
Neilsen Bookscan on the Top 10,000 computer books sold each week, and we
load that data each week into a MySQL data mart. We have nearly 20,000
unique books in the database — computer books published over the last four
years. And of those 20,000, only about 300 are currently selling at rates
that will give total sales of at least 10,000 copies a year. Granted, some
of these books will sell for many years, but others will need to be revised
annually. In short, most computer books published sell far fewer than
5-10,000 copies, and publishers are playing the averages. For every 100,000
copy bestseller, there is the 1000 copy failure, with profitability usually
found in the 5-10,000 copy range …

… And if you just say, "well screw the retailers, I’ll just sell
direct," you may only be able to sell direct — not just because they might
hold it against you, but because you’ll have skimmed the cream off sales,
and thus the modeling that retailers do of sell through when a book is first
introduced may lead them to purchase far fewer copies than they ought to,
thus reducing total sales for the book …

… Perhaps, as David says, the alternative is to bypass the entire
system, and just go to self-publishing. He uses as an example the success of
37signals’ recently published PDF-only book, Getting Real, which he
describes as a niche book.

Sorry, David. That isn’t a niche seller. That’s a bestseller. And why not?
37signals is one of the most celebrated small companies on the net, one
whose development practices are widely admired and emulated. You guys are
famous! Of course you can sell lots of copies of a book that promises to
tell people the secrets of your success.

But this doesn’t mean that every author can expect the same results, any
more than the next blogger can expect to get the same traffic as BoingBoing
or Engadget. Doing everything right is necessary, but not sufficient for
success. Some newcomers hit it just right and go to the top of the heap, but
most live out their life somewhere in the tail…"

Make sure you read the rest of it as I’ve only paraphrased the article and
comments.

Certainly I’ve had one or two offers through from publishers to do a book on
Oracle business intelligence, but I have to agree that the figures just don’t
add up, not unless you’re someone like Tom Kyte, Jonathan Lewis or Steven
Feuernstein (or Kevin Loney, as

this posting by Don Burleson notes). A typical first book on a subject
that’s been done before is only likely to sell 5000, say 10,000 tops and with a
royalty rate of between 10% and 12.5% of net price (remember, retailers such as
Amazon only pay 50% or less for their wholesale price from the publisher,
meaning that net price is likely to be $20 or less) you’re looking at a royalty
of between $1 and $2 per book. Now if you’re Tom and you sell tens of thousands
of copies (and rightly so, I might add) then it all works out, but if you’re
writing the book in conjunction with another author, and you sell 5,000 copies -
quite a probably outcome - you’ll be lucky to make $2,500 for what is probably a
year’s work, and that’s before you take off your initial advance and publishing
expenses such producing the book index.

Now in the old days, when the only route to getting your work published was
to write a book, this still made a lot of sense, but these days, when you can
self-publish on the Web, or
publish your book as a PDF
and sell it online, it’s not such a compelling deal - and you could make the
same amount of money, and probably reach a similar sort of audience, from
getting your work
published on
OTN. In the old days, books meant respectability and acknowledgment by
peers, but if you look at people such as Howard
Rogers,
Doug Burns,
David Aldridge,
Tim Hall (I know he’s written
some books but you know what I mean) and
Peter Scott, these are people
who’ve not written a book but clearly have contributed something back to the
community and have the respect of their peers. Also, from a consultancy
perspective, I need to have a medium amount of knowledge about a wide range of
subjects, and it just doesn’t make sense to devote up to twelve months on a
single subject such as Oracle Warehouse Builder.

Having said all that though, I can see the attraction in putting something
together, something substantial and lasting, about a subject you’ve got a lot of
knowledge of. What I can’t see the point of though is rushing out a book on a
new technology, just to be "first to market" and to try and make some money. The
money just doesn’t add up and there are far more effective ways these days to
build a reputation from writing.

Most job descriptions are !@+#$ (you fill in the word as long as it is derogatory). Take a look at a typical job description.

Administrative Assistant - Performs a variety of secretarial duties, such as typing reports and memos, maintaining computer based and paper files, answers office inquiries, and performs administrative tasks. Performs special projects of a moderate to highly skilled nature. Normally requires HS and 5 years experience.

ulled this off a website that specializes in job descriptions and it meets my criteria for typical and lousy. What it tells you is what the hiring manager expects the actual duties (responsibilities) of the applicant to be.

What I hate about this style of job description is that it doesn’t say why the person is being hired. Isn’t it possible that this person could get hired and spend all their time maintaining computer based and paper files and completely miss the boat on expectations?

A better way

I have to admit that I’ve written a lot of job descriptions based on title and responsibilities in my 20-year career but that’s history now. Now I try my best to describe jobs in terms of the desired outcomes, measurements and required skills. I admit that empowerment is a tired business term, but it still makes sense to give employees freedom to figure out the best way to get a job done. Tell them what you the world will be like if they do a good job.

An Administrative Assistant is a tricky one to apply this concept to because many of the day-to-day tasks are (seemingly) unrelated but I figured that if I’m going to blast the current practice of writing job descriptions I better show how it can be better. Here is my rewrite of the (brief) Administrative Assistant job description.

Administrative Assistant - Makes the development manager and team more productive by completing administrative tasks in a cost-effective manner.

I think this job description tells the person why they are here. For example, this description gives the assistant the freedom to discover that the most cost-effective way to arrange travel for people on my team is to use a travel agent rather than doing it themselves. They have offloaded an administrative task in a cost-effective way. Who cares if they do it themselves or how it gets done?

It’s been a week or so since my last blog posting, so I thought I’d better do
an update. I’ve been busy working away on the presentations and papers I’m doing
for Collaborate’06 in Nashville, and when I finally get down to some writing I
tend to enter "the zone", where I just ignore any emails or other distractions
and try and finish off what I’m doing. So, if you’ve dropped me an email or
posted on the forum and I’ve not got back, apologies and I’ll try and catch up
over the next few days.

I mentioned a few weeks ago that the UKOUG are running a
Business
Intelligence and Analytics Special Event in the Grange City Hotel, London,
on May 16th. Now what I couldn’t say at the time was why this event was going to
be so special, but if you read my posting last week about Oracle’s BI
announcements around Fusion, you might be interested to know that we’ve managed
to line up a whole bunch of senior Oracle speakers who are going to explain how
all this new technology is going to fit together. We’ve got a presentation by
David Pryor who’s going to explain Oracle’s strategy around Fusion, Siebel
Analytics and the current product line, whilst Morgan Russell is presenting in
the technical stream on how the BI product line-up will look going into the
future. Paul Badger is going to explain how Oracle’s CPM product line-up (EPB,
PM and DBI) will fit into the Fusion strategy, whilst Jon Ainsworth is going to
talk about how the trends in the BI marketplace will move BI from a "nice to
have" to a mainstream technology over the next few years. Finally, we’ve got a
couple of excellent customer presentations, a talk on analytics, OLAP and data
mining by Doug Cackett, and a presentation by Brian Gregory on how corporate
governance processes can be applied to CPM and business intelligence. Here’s the
current speaker line up: (agenda, PDF,
here)

• 
  "The Fusion of Technology & Applications", David Pryor, Vice President
  of Business Intelligence and Warehousing, Oracle EMEA

"Oracle’s own development innovation and recent acquisitions -
PeopleSoft, Retek, Siebel - have transformed and expanded the breadth of
Oracle’s Business Intelligence solutions. David will explain Oracle’s
strategy for the development of Business Intelligence technology for
organisations implementing analytic solutions, as well as providing a BI
foundation in Oracle Fusion applications."

• 
  "A Vision for Business Intelligence- Business Intelligence Today, the Trends
  and the Future", Jon Ainsworth, Solution Sales Director, Business
  Intelligence & Data Warehousing, Oracle EMEA

"In this presentation, Jon will review the reality of BI adoption in
business today, plus some of the business drivers that affect it going
forward. In this context Jon will then discuss some of the likely technology
and functional shifts that we are likely to observe as BI evolves into a
”main-stream technology “ over the following years."

• 
  "Integrated Architecture For Business Intelligence - Roadmap To BI Fusion",
  Morgan Russell, Director of BIW Technology, Oracle Corporation, EMEA

"Attend this presentation to understand how Oracle is moving down the
road to BI Fusion with the addition of Siebel Analytics to its BI portfolio.
This presentation will provide a product update on Oracle’s complete range
of BIW technology. See for yourself why Oracle is excited about BI."

• 
  "A Journey From Information To Knowledge", Rod Swift, Head of
  Application Development, RIAS

"The journey from Information to Knowledge can be littered with broken
intentions and compromised principles. The final destination rarely matches
those glossy pictures in the guidebook. At RIAS, we do not claim to have
discovered Utopia but we do believe that we have arrived at a BI solution
that has much to commend it. Hopefully, the sharing of our experiences may
make your journey that much richer."

• "Why OLAP & Data Mining Should Be Part Of An Integrated Process",
  Doug Cackett, Oracle Corporation (abstract TBA)
   
• "Planning, Budgeting, Monitoring & Analysing your Business", Paul
  Badger, Oracle Corporation (abstract TBA)
   
• "Agile Procurement: BT And BI", Stephen Martin, BT, Chief Architect -
  Business Intelligence and Data (abstract TBA)
   
• 
  "Can
  You Apply Corporate Governance To Your BI/CPM Data And Process?", Brian
  Gregory, Senior Director, Compliance, Oracle Corporation EME

"A great deal of attention and effort has been correctly applied to
ensuring that transactions in a compamy’s ERP system are correctly
controlled and recorded. But can management say the same about their CPM/BI
data? If they can’t what is the value of this data? This session will look
at the challenges and how Oracle’s Performance Foundation can help you apply
Governance principles to CPM/BI data creation."

There’s also a partner stream where the company I work for, and Noetix, will
get to do a presentation. As long as I don’t have to go up head-to-head with
Morgan I might even do something on how the new Oracle Analytic Server works,
but it depends on whether I get enough free time between now and then to put
something together. Until then, you can
register for a
place here (it’s free, even for non-members) and I’ll hope to see you all on
the day.

In all of my working life I have never lived and worked in the same city, true I always tried to minimise my daily commute but working in the computer service industry there is always going to be some travel. But this will change in the summer when my office relocates to a five minute drive (or more likely a 15 minute bike ride from my house) I could even walk to work!The only downside to the move

We recently worked with a customer on a Live Server application that pulled data into Excel from SQL Server, SQL Server Analysis Services, Oracle, ProClarity Analytics Server, and a web site. After we finished the application and cleared the sweat from our brow (integrating data from that many sources has it challenges), we again recognized that the true challenge with any application is the data itself.

When we demonstrated the application to a couple of IT folks, we explained that we couldn’t display exactly what they wanted because their data had no common keys or names. There was simply no way to map the data from one source to the related data in another.

The IT guy—who probably hadn’t been “in the trenches” for several years—was incredulous, stating with an edge of irritation in his voice, “we’ve had this problem for the past five years!” (And just to be clear—he wasn’t irritated with us. In fact, I think he quite liked us. He was irritated that the same old data problems were the same old data problems.)

It was a bit of an “ah-ha” moment for me (which always occurs when I get outside of the lab and actually work with customers) because we had helped them take a small step in the right direction. As we began developing their application, we pulled several of their “chaos” data sources into a database. One of these was a customer list that was manually created and maintained in Excel because they felt the official customer list from the CRM system was too hard to sort through.

Then, the magic happened. We created a writeback data request for the customer list, and without disturbing any of their existing processes, we gently moved them away from disconnected, unmanageable data to centralized, secure data—data that could eventually be incorporated into their backend ETL processes. Although nothing had changed for the Excel users (they were still able to see and maintain their subset of customers in Excel), something significant had changed in the organization’s ability to capitalize on their data assets. This data was now being stored, managed, and updated in a central database.

The change may extend to standardizing customer names and keys across all data sources, or it may simply be a better way to manage customer contact information. It wasn’t a dramatic “we’ve-revolutionized-the-way-we-do-business” change; it was just a tiny adjustment on a corporate gear to make a small something run more smoothly.

Let’s say you had no (easy) way to patch your product and deal with individual customer’s problems once your product is released. Then let’s say you are going to ship millions of copies in the first few months after release. How would you change your approach to software development?

Read the following article on Zero-Defect Software Development to give you an idea what ZDSD is all about.

In this article, Steve Pavlina outlines ten basic rules to follow.

• Test your product every day as you develop it and fix the defects you find
• Review your code regularly
• Rewrite poor-quality modules
• Assume full responsibility for every bug
• Handle change effectively
• Rewrite all prototyping code from scratch
• Set QA objectives at the beginning of every project
• Don’t rush debugging work
• Treat the quality of your code at the same level of importance as the quality of your product
• Learn from every bug; each one represents a mistake you made

I really like this article because it struck a chord on my personal love/hate relationship with QA. I love QA (and particularly our QA people!) because they help us create a much higher quality product than we would otherwise have. However, I hate the tendency of our organization and other software development organization to rely on QA people to “create” quality in software products. Manual, ad-hoc, exploratory, whatever you want to call it is a necessary evil. Using a quality approach from the front end is the only way to effectively create a quality product. The traditional, test it when you plan to ship it, approach will always lead to an ever increasing load of deferred defects.

However, what I see missing in Steve Pavlina’s list is perceived quality. His approach is solid but it completely misses the fact that quality is in the eye of the beholder. You could do all the things he lists above and completely miss the mark with the customer. I contend that you can’t achieve the customer’s concept of quality until you truly understand and empathize with the customer. For example, if you implement 9 out of 10 features necessary for the customer to complete an important task have you done your job? The answer is NO. Most software dev organizations would take the request for the 10th feature as “a feature request” and not a quality problem. If it showed up in the defect tracking system it would likely get categorized as “Works as Designed - Not to be fixed”.

The problem is related to the definition of a bug. Most organizations consider a bug to be any gap between how the product operates and how it was designed. I define a bug to be any gap between what the customer expects your product to do and what it actually does (assuming the customer is using your product to solve the targeted problem - if they are using a blow torch to cook a steak it may not be a bug). I told some key developers this in my organization in front of managers of our support organization and I got some wide eyes. The question came up “Does that mean a feature request can be escalated for a hot fix?” I said yes, feature requests should be candidates for hot fixes. If we miss the mark on requirements we should own up to the problem.

The bottom line is that I believe the first and most important step in achieving quality is getting the requirements right.

Business Intelligence Pipeline | Q&A: Process Management Grows Up: “Intelligent Enterprise (IE): Can you recap the maturity model you unveiled at the Gartner BPM Summit?

Jim Sinur (JS): There are five stages: process understanding, process control, enterprise execution, corporate performance management and, last, competitive differentiation. You start modeling and measuring in the process understanding stage. In the second stage, you’re doing more optimizing and tweaking. This is where you take advantage of rules management and optimization. In the third stage, you craft a cross-enterprise process architecture and maybe extend that to your trading partners. In the fourth stage, you instrument a framework of corporate performance management goals against the actual process.

IE: How mature are the companies you encounter?

JS: Most are in stages one and two. Basically you move to larger and larger scopes [of deployment] as you move up the maturity curve and to more dynamic competition. Once you get to stage five, you’re using your processes as either a defensive or offensive weapon. ”

I came to the BI market without much knowledge about Alphablox before IBM acquired them. So I did a little bit of digging using my best friendGoogle,

Founded in 1996, Alphablox is headquartered in Mountain View, California, and has offices throughout the United States, the United Kingdom, and Australia. Alphablox is a recognized innovator and was selected by Computerworld Magazine as a “Top Emerging Company for 2002″; by Intelligent Enterprise Magazine as “A Company to Watch in 2002″; by Upside Magazine as “A Hot 100″; and by Technologic Partners as one of ten “Investors Choice for 2002″ award winners.

Officers: CEO, Joseph Guglielmi (w/03); Technology, Bill Wagstaff (w/04); Sales/Marketing, Steve Guttman and Michael Nason, Original Alphablox Office: 516 Clyde Avenue Mountain View, CA 94043 (Source)

The strength of Alphablox lies in its architecture. It is the first 100% Web Based Custom Analytics tool. During its time, the BI Market leaders like Cognos and Microstrategy develop powerful client based BI Solutions. These are powerful and useful tools made available to different Fortune “rich” companies.

Other BI solution providers forgot that these ‘rich’ companies started doing OUTSOURCING. Their HR, Finance Dept, or Helpdesk Support might not be in a central data warehouse nor country/time zone. These ‘rich’ companies also started experimenting with co-locating part of their business in other countries. They also made experiments with the corporate structure and hierarchy, specially with empowering the decision makers. The right thing that Alphablox did that time, is that it did not compete with these great and successful companies, they develop a solution ahead of its time by being “100% web-based” (no need to install anything to access data), custom BI tool (made it robust and modular - use only the components that you need), flexible on how it will be deployed and extensible up to the level of interoperability with other industry tools.

With that strength in mind, they were acquired by IBM on July 2004 and was later repackaged to be part of the IBM DB2 Enterprise Edition.

Client Focus: Pfizer on 2001

From the maker’s of Zyrtec®, Zoloft®, and Viagra® says they’ve tried multiple BI products that couldn’t be customized, and they were forced to use proprietary infrastructure. The learning curve was high, yet the tools lacked important features and functions. Thye wanted to build applications customized to users, not force their users to customize themselves to some packaged tool.

“The technology worked right out of the box. Just drag and drop onto the templates, and you’re up and running. And once the sales organization saw the prototypes, they got excited about what we were doing.”

“The beauty of AlphaBlox is that you don’t have to hire a bunch of expensive consultants for months just to implement it. Anyone with knowledge of HTML and JavaScript get can started building applications right away.”

“We chose AlphaBlox because it gave us total control over the application look and feel, as well as the ability to easily integrate with other sources of information. All we have to do is set one attribute, and we can completely personalize the interface. “

“It’s not about the database or the schema—it’s about information delivery.”

“If people can’t access information and transform it into knowledge, then the technology is meaningless. AlphaBlox makes it easy to rapidly assemble powerful, easy-to-use analytical applications that address the full gamut of requirements.”

Related Articles:
1. What is DB2 Alphablox
2. 10 technical things every DB2 Alphablox Developer should know
3. Blox Team Pics and Why we choose Alphablox