geordee


Only One Miracle At A Time

Over-engineering is something every designer, every architect should be beware about. In the essay The Second-system Effect, Frederick Brooks points out one common pitfall - when the architect designs his second system, it is the most dangerous system he will design, because he will try to incorporate everything that he missed in the first system. I have observed this to be very true in many occasions. Over a period, architects get out of this. Slowly they tend to build balanced, and mostly conservative systems as they age in the profession.

I use a principle while building systems - Information Technology...


It's The Data, Stupid!

We started our efforts to build software products a few months back. Soon we churned out some small applications, as part of our training program. The product is a small library management system, built on Ruby on Rails, the open source way. Soon we started getting enquiries, from big and small libraries who are on the way to computerization.

The product’s evolution is from a training program. It boasts nothing that refers to Z39.50, MARC-21 or anything related to libraries. In fact, the key concept around which the product was rental inventory management, making it customizable to track a generic...


Single Version Of The Truth?

One of the often-quoted phrase, almost a cliche, in Information Management space is single version of the truth. In the past I have quoted to justify or sell master data management and enterprise data warehousing solutions. Off late, I have started looking objectively on the business necessities of single version of truth. Partly, this is due to the observation that not all business scenarios require a single version of the truth.

Is it because the IT systems or databases are unreliable or incapable of defining the truth or facts? No. It is due to many different reasons. While the truth...


Rethinking Business Intelligence

The only constant is change is ever so true with Information Technology. IT systems and solutions are created in the context of the evolutionary period in IT resources. From a number of implementations we derive technical know-how, patterns, best practices, and wisdom. Sometimes the evolutionary cycles take us to newer possibilities that invalidate the current common wisdom. So is with business intelligence.

It has been almost two years since I wrote the demise and restoration of data warehousing. The brave new world is now established.

Business Intelligence aided by data warehousing...


Crossing the Golden Gate

I have been playing around with GoldenGate for a few months now. The more challenges I face, the more I like the features and capabilities of this tool. Except for a few bugs introduced in the recent versions (typical in the Oracle products I have worked on), GoldenGate proved a solid replication and changed data capture tool. Now, “solid” should not be understood as rigid; GoldenGate is quite flexible in the way it can be configured.

One of the challenges we had is to extract data with minimal burden to the source databases (in Oracle). The source database was configured...


Oracle Spend Classification

As information technology progresses, computers become more capable to perform activities which requires not just basic binary logic, but approximation, correlation and prediction. Such implementations help us teach computers to perform large and complex tasks which otherwise require an intelligent supervision. In common technological parlance, this is known as machine learning or knowledge discovery. In relation with data warehousing, we know this as data mining, which is a prominent technique for knowledge discovery.

In large organizations controlling expenses is as important as making profits. Procurement and Spend Analytics is therefore an important area for analysis and business intelligence in large...


Encoded Images in OBIEE 11g Report

Business Intelligence tools are not omnipotent. Different tools have different capabilities, and once the tool selection is over we generally live with what the tool can offer. In this context, I often suggest that the requirements gathering can be done in a tools context, both to ensure we can achieve what we want and to best utilize the tools capabilities. Once in a while, clients express a requirement that is just beyond the tools capabilities and we wish we could somehow push the boundaries a little.

Recently one of the clients asked whether they can display thumbnail images along with...


Data Warehousing is Dead, Long Live Data Warehousing

Data Warehousing has turned over a new leaf. Analysts are thinking beyond databases and ETL tools. Vendors are talking loud about their Big Data solutions, Hadoop integration and NoSQL products more than ever.

The old data warehouse understood our data through metadata. The old reported the patterns and trends we have asked. The old helped us find answers to our questions. The old just talked. And now, the old is dead. Data warehousing is dead.

Long live data warehousing! The new will make us understand our data. The new will go beyond dimensions and facts. The new will do more...


From Data to Mining

The Internet is abuzz with Big Data, Hadoop, NoSQL, petabytes and portmanteau words starting with Exa, of course only if you care about databases and data warehousing. It was all good, until the hype hit, and surely it has hit. We are now trying to fit in available big data processing techniques everywhere. How do you store big data? How do you process big data? How do you crunch big data? Hardly, I hear questions like what do you do with big data? We hope that the hows will lead us to the whats. Sort of means justify the ends...


Expanding Universe of BI

Recently, we had a discussion on how the future of BI is shaping up. As I was exploring the trends in business intelligence, it seemed the new developments are not in any particular direction. Today business intelligence is expanding in all directions. The presentation aspects of BI have already embraced rich UI, Flash and Ajax. It is now reaching out to HTML5 and mobile platforms. There is also a trend towards data visualization, rather than traditional tabular representations. The analytical aspects of BI are now developing towards massively parallel platforms - be it MPP databases and appliances or frameworks like...