Welcome to my second business intelligence (BI) column. I`d like to share my thoughts about a component that is often neglected when making an intelligent BI decision.
I`m talking about scalability. As with many other enabling processes, organisations need to bear future systems growth in mind, but should insist that their choice of software achieves the performance requirements across the enterprise as a whole.
At the heart of the matter is the fact that "scalability," as it applies to BI, is poorly defined and understood, by both vendors and customers alike.
Vivek Thomas, Guest Columnist, ITWeb
As a company`s demand for BI tools expands throughout the enterprise - potentially bringing together thousands of users in one integrated system - the increasing strain on product scalability can be considerable, often to the point of system failure.
Indeed, system scalability has become one of the primary performance concepts for evaluation among analysts of the BI industry. Not surprisingly, virtually all BI vendors have responded by making loud and frequent claims of scalability. And that is a problem because very few of them are scalable.
At the heart of the matter is the fact that "scalability," as it applies to BI, is poorly defined and understood, by both vendors and customers alike.
Scalability embodies a combination of complex technologies and concepts, many of which may not be readily apparent at the time of system evaluation and purchase, but are destined to surface eventually. Much like an iceberg, scalability appears easily defined by very obvious considerations on the surface - but a whole range of requirements lurks below.
Failure to understand and evaluate product performance at any of these levels may jeopardise the entire system and increase the risk that the selected system will not meet an organisation`s long-term information needs.
Increasing number of users
As organisations grow, the number of users in a modern organisation consequently grows. Whether this growth involves scaling from a LAN to a WAN, or a worldwide intranet or extranet, this reality must be addressed to allow for medium to large-sized implementation.
More than just facilitating growth by making incremental hardware additions, a scalable BI solution should provide some form of intelligent user management. This includes the ability to effectively load balance work among employee`s tasks, the ability to prioritise among the user base, and the functionality to ensure each information user has a secure and flexible view of the corporate data.
Increasing number of objects
Concurrent with the increase in the number of users is the increased demand for numerous flexible objects, such as reports, queries and cubes. The addition of new users will result in several important demands on the system: increased viewing demands for existing objects, new user requirements for information manipulation after it is initially received, and demand for creation of new productive objects. A robust and scalable system must be able to gracefully handle these increased and varied demands.
The ability to process objects on a dedicated tier or machine is crucial to increasing the load. The system`s architecture needs to support thousands of objects and manage the resulting output and files.
The system should also provide the intelligence to minimise object duplication to aid in distributing them efficiently throughout the organisation.
Increasing amount of data
Another consequence of a growing organisation is the requirement to manage an expanding quantity of corporate data. As companies increase their ability to gather data and expand their range of storage options, their BI system must scale accordingly. A scalable BI solution should be able to accommodate ever-increasing amounts of data, as well as extract it from varied databases. Furthermore, the system should process reports and programs intelligently so that the involved data sets sit on the most efficient tier of the BI architecture, thus minimising the data that is presented to the client.
As data warehouses become increasingly large in size, it becomes critical that software is smarter about where it aggregates and stores data.
Increasing numbers of transactions
As the BI system is deployed to ever-greater numbers of users in the organisation, the sheer volume of transactions - requests of the database, scheduling of reports and queries, distribution of information to key people - will increase dramatically. Sharing valuable information among decision-makers is at the heart of BI. The ability of a specific vendor solution to efficiently handle large and consistent increases is a fundamental component of scalability.
Functionality
Scalability also needs to be addressed in terms of functionality. As I see it, most BI products hit a wall in terms of functionality, as they are unable to cope with the customer`s more complex requirements.
If a client is looking to get the most out of its BI solution over a long period of time, it is best to ensure it implements tools that have a common underlying architecture and therefore a large degree of crossover in terms of functionality.
Reliability
Before I wrap up this month`s column, I`d like to add some comment about what seems to be an obvious point, that of reliability. With users wanting to access the information residing in the various repositories of data throughout their organisation, they need to know that the information is actually there and accessible. When implementing a BI solution, IS managers need to make sure systems are in place to ensure the server that delivers the information is up and running. In today`s business environment, it is critical for users to know if things have gone wrong.
The scalability of a BI system is a critical point of evaluation for IT professionals. Conducting such an evaluation represents a challenge, as the concept is poorly understood in the market. Indeed, most vendors prefer it this way, for while most make frequent and loud claims of product scalability, few actually provide it for their customers.
Given the mission-critical nature of the BI deployment and its associated costs, it is imperative that IT professionals understand what true scalability encompasses. They should ask prospective vendors tough, challenging questions to ensure they select a product that will advance their business - not only today, but for the long-term.
I would like to stress that there are a number of additional issues that may be discussed around the topic of scalability and if you are interested in continuing this discussion please e-mail me at vivek.thomas@seagatesoftware.com.

