Most IT spent in the past 25 years has been on ERP and communications. With the former, companies' reasoning has been that it would give them three things - a central repository; streamlined and integrated processes: and a means to analyse the business.
While it has been successful at the first two, it has failed at the third. ERP and databases are good at storing and collecting data - not at reporting on it.
That's where business intelligence (BI) came in. It uses online analytical processing (OLAP - the building of cubes to slice and dice data to answer a query) and reporting tools to address this need.
But BI has been widely criticised for the way in which it achieves these objectives, and in truth one can arrive at the same goals in a far more elegant way.
PROBLEMATIC REPORTS
Printed reports, however, still remain the primary mechanism to analyse data. The problem with reports is that they work with "dead", non-associative data. They are drawn in pre-defined formats with pre-defined parameters. Once you get a report, you cannot re-arrange it into another view.
The most obvious drawback is waste - of time and paper, with interminable reports becoming necessary, and in over-reliance to the IT department. It's a time-consuming, expensive, inefficient way of arriving at the right information.
One need only think of the volumes of reports necessary in a multinational with the daunting product catalogue to see the need to go beyond BI.
The other issue with traditional BI is that it is a disk-based technology due to the inherent hardware limitations that existed at the time of its development.
|
This is why OLAP-based BI solutions require lengthy and restrictive processes of pre-definition and pre-aggregation of data hierarchies.
HOW THE FROG LEAPS
A new technology called Associative Query Logic (AQL) has now emerged. Apart from the fact that it offers an innovative user interface, it is designed to operate in memory (RAM).
QlikView's tools, which use the patented path of AQL, fetch data from whatever source, as long as there is database connectivity, bring it into a data model and create associations between data, based on similarly-named fields of data in the source.
The upshot of this is that there is no need to define complex hierarchies, no cubes need to be created and the process of creating the data model is simpler and faster. Since AQL is designed for instant in-memory manipulation of large data sets, any analysis is done in memory and the accumulation of data is done on the fly. AQL provides very effective compression algorithms to optimize memory usage.
64-BIT WINS
The big win here is that applications that take advantage of AQL can also take full advantage of the 64-bit world, which has no practical constraints on memory.
Tools such as QlikView have simplified user interfaces that enable all users in an organisation, from the MD to the stock controller, to easily access the information they need. The powerful data access of some of these tools could in fact make it somewhat dangerous for users other than top-level management to operate them, which is why password protection is a given.
|
