Infosys Labs


BI in a Cloud: Defining the Architecture for Quick Wins

By Bharat Chadha and Meena Iyer

Cloud platform can accentuate the benefits of BI and add more value to its business reason

The past decade has seen rapid evolution of the business landscape and business organizations are increasingly realizing the need for a more scalable and flexible information technology architecture. The ever mounting burden of regulation and compliance is further amplifying business expectations from IT while at the same time tightening the noose.

However, the IT industry has been quick in its response with innovation in technologies such as virtualization, SOA, Web 2.0, etc., thus strengthening its position as a strategic business partner.

In the wake of the present economic crisis and the pursuant business environment, IT has further concretized its strategic relationship with business with the re-introduction of grid technology in the form of cloud computing. The cloud model enabled by SOA provides flexibility and scalability (infinite in certain cases) using external computing and processing power in the form of real time e-services. The primary benefits driven by this model are business agility with lower costs, enabling organizations to respond quickly and effectively to the ever changing business environment.

This, according to the IT pundits, is the future of computing.


BI in the Cloud

The cloud model requires leveraging the traditional on-premise enterprise architecture by enabling higher economies of scale. Having realized this need, many organizations (IT and non-IT) have already begun democratizing their globalized IT landscapes, to offer as services both internally and externally. These services that are ubiquitous in provisioning are termed as utility computing services as consumers pay in consonance with their utilization. This service has further resulted in burgeoning growth of ancillary service lines viz., SaaS, IaaS and PaaS.

In recent times, business intelligence (BI) has been under mounting pressure to evolve as an all pervasive information and analytics agent. The strengths of the cloud model viz., accelerated speed-to-market, reduced TCO, scalability, etc., has led many BI vendors to introduce cloud services as a clear and distinctive extension to the on-premise and on-demand BI applications [Fig. 1].

Companies like Amazon offer unlimited processing power and storage thus allowing any business to cater to its increasing information pile while keeping the IT related costs under control. On the other hand, vendors like SAP BusinessObjects provide businesses on on-demand platforms to host client warehouses and databases. There are many such related strategies and vendor options available with customers to ensure security and confidentiality of their data. To cater to on-demand reporting and dashboarding, vendors such as SAP BusinessObjects are enabling on-the-move users to self-service, thus reducing dependency on IT and increasing the scope of BI in the organization.


Typical BI Architecture (Present State)

The amount of data at the disposal of global organizations is greater than ever before and is growing at a rate unsustainable for the present IT budgets. These global organizations are getting to be more and more dependent on BI to not only enhance their decision making capabilities but also to increase the intrinsic value of their business processes and ROI from the present IT landscape. A typical BI architecture may involve heterogeneous data models for different geographies, business units, etc., depending on the information flow of an organization [Fig. 2]. This poses a problem for BI architects as scalability would entail extra expenditure in hardware and software, more complexity for data fungibility and loss in data quality and consistency, amongst other critical issues.

Many organizations are, thus, slowly complementing the decentralized architecture with a SOA enabled federated data warehouse solution, to satisfy the increase in demand for collaborative and globalized decision making. This movement, towards Enterprise Information Integration (EII), has been further accentuated by the emergence of the cloud. Gartner, in a recent research, reported that many organizations are increasingly adopting cloud based application services even in midst of fears over loss of data control, security and privacy [1].

In the current business environment, tightening regulations with economic uncertainties have further added to the need for business agility. Gartner reports that given the perceived complexity of BI, just about 15-20% of business users use BI. The same report says that by 2012, emerging technologies will make it easier to build and consume analytical applications thus marginalizing the role of IT in BI [2]. In line with this forecast, some of the reporting and data analytics service offerings in the cloud are fast picking up steam for offering self-service capabilities to 'on-the-move' business users.

Though the cloud can augment an organization's existing BI capabilities at a marginal cost, IT departments still need to stress test their BI architecture to decide which service suits them best. The criteria for this stress test have to evolve from the business landscape of the organization and not IT alone. Once these criteria have been listed, extreme use-cases have to be tested against the present capabilities to relate which cloud features can fit where. An example of such a stress test is given in Table 1 overleaf.

Stress Test Yout BI ArchitectureCloud Challengers and Mitigation Strategy 


Once the BI/IT architects along with business process owners have evaluated the BI architecture and identified the spaces/business requirements capable of cloud fitment, the next step is that of redefining the BI architecture for consistent functioning of on-premise and cloud-hosted information systems.

The architect has an option of transforming the complete architecture into a cloud model but it is not advisable considering that BI as a service is still in a very early stage of adoption. On the other hand, a hybrid architecture consisting of both on-premise and off-premise systems will be more beneficial as it will not only cater to the business requirements but also help in countering some of the challenges and concerns raised around the cloud model.

The data staging or inbound layer forms the most granular stratum of the data model. This layer that also houses the extraction programs is the most frequently impacted one despite efficient loading schedules. Thus it requires the maximum amount of housekeeping by the system administrators. This layer is the most susceptible one to come under the BI architect's radar to transform onto a cloud service. By pooling resources from various geographies/business units, enterprise architects can form a private cloud by centralizing the computing power and capacity while virtually partitioning the allocation of these resources. The BI architect can then host the staging layer on the cloud by initiating an appropriate data governance model at the data source layer itself and utilize the extraction logic to cleanse and profile the data [Fig. 3].

Another layer from the BI value chain that is most likely to be hosted on a cloud is reporting and analytics [Fig. 4]. Decision making in multi-national organizations is evolving towards a collaborative model with diverse business users performing data mining, intuitive search, etc., on enterprise-wide data to make decisions of global effect. BI architects can create a cloud using internal resources to host reports from various business streams. They can also host these reports using a public cloud for seamless access by their mobile workforce, provided an appropriate data governance model is introduced at this level as well.

Data Staging on a Cloud – A Federated ApproachReporting and analytics on A Cloud – Eabling Self-service 


Thus it is imperative to understand that BI as a service offers huge possibilities for removing barriers (e.g., geographical) to decision making by integrating high volume and mission-critical business processes. Irrespective of the age of a BI landscape the cloud model can drive increased BI adoption, improved end-user experience, better access to analytics and reduced IT dependence.

In the world of economics, a production possibility frontier (PPF) graph is frequently used to determine the market and non-market trade-offs to ensure optimum usage of technology and resources. Similarly, an organization will have to evaluate the present enterprise landscape to optimally allocate the available resources towards the ever increasing BI requirements. The need of the hour is for a scalable and robust architecture reducing TCO to make trade-off decisions much easier to implement.


  • Data in the Cloud: Changing Nature of Managing Data Accessibility, Gartner RAS Core Research, February 2009. Available at
  • Gartner Says Emerging Technologies Will Marginalize IT's Role in Business Intelligence, Gartner, March 2008. Available at
  • Strategic Alignment, Predictable Performance, and Confident Decisions, SAP BusinessObjects Portfolio.
  • Thomas H Davenport and Jeanne G Harris, Competing on Analytics: The New Science of Winning, Harvard Business School Press, 2007
  • Rahul Kulkarni and Maxim Sychevskiy, EDWH Architecture for Global Data Loading Strategy, SAP Community Network, 2009
  • Booz Allen Hamilton, Enterprise Architecture and Cloud Computing, 2009
  • Federated Data Warehouse Architecture. Available at

Share the page

  • Tweet
  • Share on Facebook
  • Share on LinkedIn
  • Google+
Back to Top