Business data is one of the most critical components of the IT portfolio of any enterprise. Most e-business applications are responsible for reading and writing business data in some form or other. Therefore, the efficient storage, retrieval, and management of the data constitute a challenging problem in all organizations.

Enterprises with multiple lines of business spread over multiple geographies have critical data stored in multiple, scattered databases. In many cases, the scatter of core data is proportional to the size of the enterprise IT portfolio. Organizational growth, such as mergers and acquisitions, compound this problem. Such companies may have heterogeneous data environments with varied schemas and they may contain redundant data elements. This data may be static reference data, such as personal customer information or geographical data, common business data, or common external data such as market data. This can lead to serious inefficiencies and consequently higher costs because of the overhead in accessing/updating data in multiple databases using different mechanisms. These issues lead to an incomplete view of core business data such as customer information that can cause inconsistent user experience and the introduction of risk. Apart from this, interoperability between business lines is often difficult and error-prone. In this paper, the overall case for shared data services is developed with a reference implementation based on financial services. In a future work, this approach will be applied to a different industry vertical and subsequently, specific implementation concerns will be addressed along with migration strategies.

Business Problem
Consider an example from financial services where some of the larger players have multiple lines of business, such as personal banking, personal insurance, and credit cards apart from scattered geographical presence. Common data, such as customer personal data, may be maintained in multiple repositories and checked for consistency in separate applications using proprietary logic. This can lead to issues in synchronization, user experience, etc. A typical problem that can occur is related to a portal through which a customer may update his or her personal information through one LOB, such as personal banking. Due to synchronization lag, this data may not percolate to the database associated with other business applications, so the customer may see outdated information in other LOB applications such as personal insurance. The key issue is that there is no holistic view of core customer data. This can lead to issues in satisfying business requirements. Because of this, other applications such as CRM, risk management, etc., do not have a comprehensive and current view of core business data. Figure 1 illustrates a detailed business context where this situation is applicable.

Note that this represents a typical financial services organization with offerings in the insurance, asset management, and brokerage domains. In addition to this, the company has also expanded geographically to provide insurance offerings in Europe. A multitude of applications access and update critical data to multiple databases in each separate line of business and geography. Dynamic business needs and the push from competition have created drivers for cross-business integration and the desire to provide users with an integrated experience. This in turn has led to increased demands for the availability of core data from other lines of business and the need for the data to be in sync. Currently, this is performed by using custom synchronization routines that are unstable and difficult to maintain. Apart from this, these types of solutions are not scalable given the expanding nature of business, the nature of different business units, and the technology platform that may be in use in multiple lines of businesses. In addition to the Web applications, note that there is a team of dedicated back-office representatives who interact directly with customers to resolve various issues, set up, etc., and directly update databases through a suite of desktop applications. An IVR channel also exists through which customers may directly make changes to core personal data. These update channels present an additional challenge to the integration of data and the creation of consistent views of core corporate data such as customer reference data.

To summarize the key challenges of this problem domain:

• Critical business data is located in multiple repositories
• Data has multiple channels of update, including "back door" updates
• Data needs to be synchronized between repositories
• Needs and usage patterns of data are diverse across systems
• Different lines of business have different technologies, hence it is difficult to create a common data dissemination strategy
• Performance requirements of data access are increasing due to the evolving nature of business
• In the long term, it is more beneficial to move away from siloed data models from the cost and agility perspective

Based on this problem statement, let's examine some options for tackling these problems.

Solution Tracks
Based on an analysis of the problem statement above, one can propose two parallel tracks of activities that can mitigate the problems mentioned above over a period of time. Let's examine the tracks below and later we'll look at a specific track in additional detail.

Develop a comprehensive DSA (Data Services Architecture)
In order to develop a DSA, we need to focus on the following areas, keeping in mind the specific criteria that need to be addressed in each area.

• Integration and consolidation of data
  - Work with the technology and business groups to develop unified schema for common data
  - Identify target databases for storage of common data
  - Develop a coherent migration strategy to attain the new DSA
• Dissemination of data
  - Needs and usage patterns of data are diverse across heterogeneous consumers
  - Develop a scalable strategy that can accommodate new applications with minimal turnaround time
  - Ensure that data consumer applications are insulated from issues with data management

Rationalization of processes

• Data has multiple channels of update, including "back door" updates. This needs to be examined and modified to make sure that updates are performed consistently.
• Develop a strategy that will move away from siloed data models in the long term.

Data Services Architecture
In this article, I will focus on the data services architecture. The other aspects mentioned above are very important and can fill up journals on their own. Specifically, issues such as the development of a common data model can be very challenging from the organizational perspective, and much more so than from the technical standpoint. I will focus instead on the architecture that can provide a viable option for integration and dissemination of the data given the discussion in the solution tracks section discussed above.

• Figure 2