One of the key issues before enterprises today is managing data explosion.
And business intelligence (BI) software plays an increasingly important role in
building fast and scalable databases to service the complex business needs of
such organizations. Andrew Rouche, head, systems, Web server division, IBM
Asia-Pacific, has worked on specialized practice-delivering systems
administration, consulting and services for major corporations deploying
Rouche was recently in India for the launch of IBM’s NUMA-Q range of
e-servers, which benefit BI. He spoke to DATAQUEST on BI and how NUMA-Q
architecture benefited companies. Excerpts:
On the key requirements for a knowledge-hungry company:
Modern businesses have learned that they are more effective when the
decision-making power for particular issues are in the hands of the employees
closest to them. Thirty years ago less than 10% of a company’s employees made
important decisions. Imagine that kind of scenario in modern business! It has
reversed completely, with around 90% of the employees being decision-makers now.
This means huge data flowing in and out of the organization, and to tackle this
problem, a modern, knowledge-hungry company needs a BI solution.
On how BI solves the needs of an enterprise:
BI solutions help decision-makers manage data effectively, and this is a
challenging task, since the amount of data that can be captured is exploding.
The number of people within a company who need access to the data, and the
number of different ways in which data can be accessed and used, are matters of
concern too. This calls for an economically and efficiently scalable BI system.
On the prerequisites of BI architecture:
With the Internet access a norm rather than an exception, users of a BI
system may be located anywhere in the world. This necessitates constant uptime
with continuous access, and hence the scalable model. A plethora of requirements
must be accommodated in one overall solution that allows inter-operability.
There is also a growing integration of separate components of a BI solution;
hence the need for flexibility.
As companies grow, they normally acquire customers spread over diverse
geographies. With the Internet connectivity increasing worldwide–which demands
a growing selection of application software–raw data is being made available
abundantly. The ultimate purpose of any BI solution is to transform these data
into usable knowledge. A good example is supply chain management (SCM) where
data captured from customer purchase transactions is fed into programs that
automatically configure orders to suppliers. So, it analyzes customer behavior,
projects future demand for products, reviews current inventory and adjusts
purchasing pattern accordingly.
On how BI can reduce total cost of ownership:
Considering that the investment in software and hardware represented by BI is
on the higher side, the investment for training personnel may be greater. Here
comes the manageability of the BI system. If it is simplified, it allows the
company to reduce the number of highly paid specialists who are retained to
administer and maintain it. So, I think manageability goes a long way in
reducing total cost of ownership.
On the NUMA-Q range of servers:
Ten years ago, it would have been hard to find a business database as large
as a terabyte. Today, that is a common phenomenon among corporate users. IBM’s
"terabyte club" (customers running databases of at least a terabyte of
storage on IBM systems) has more than 170 members, and recently an IBM
NUMA-Q2000 server was configured for a proof-of-concept system totaling 82 TB.
IBM NUMA-Q architecture forms a scalable foundation to keep pace with the
unpredictable growth in data. It follows a modular "quad
architecture", which allows easy, flexible and balanced growth, and all
quads are tied together with NUMA-Q’s IQ-Link interconnect technology. Each
quad is an Intel-based processing package comprising four processors, local
memory and multiple I/O ports. Expanding an NUMA-Q system is easy, unlike system
expansions under massively parallel processor architectures. Adding performance
to an NUMA-Q system does not require reconfiguring the database across hundreds
of MPP nodes, with days of down time. In fact, a company can plug in enough
quads to double its system capacity and the operating system will detect and
configure the new quads automatically.
On global companies, that has deployed BI on NUMA-Q:
A good example is the Ford Motor Company that has built a global warranty
system based on the NUMA-Q datawarehouse. The strategy of turning the world’s
second largest automobile manufacturer into a global organization involved a lot
of reengineering in its manufacturing processes. Now the company has a system
built on a corporate-wide datawarehouse that stores vehicle information and
warranty claims. As each vehicle is manufactured, detailed information on its
exact specifications is captured in the database. The consolidated information
allows Ford engineering and manufacturing organizations, in conjunction with the
company’s suppliers to rapidly identify emerging product quality problems and
to correct them. Credit Lyonnais, National Security Association of Software and
New Westminster Bank are amongst our other customers.