It`s a simple fact of business life in SA and the world that large corporations have silos of data and applications restricted to particular departments and divisions in various locations. The way businesses operated in the past made it acceptable for departments to buy their own IT as long as they performed their functions efficiently and stayed within their budget.
With companies now globalising, more executives are realising the importance of data and the necessity to make it available across departmental boundaries. The traditional solution is to develop data marts or data warehouses in which information from diverse sources is collected, cleansed and made available for analysis and querying to a broad range of users. A benefit of this approach is that operational databases are not affected by analytic applications, no matter how data-intensive they may be.
Yet, even in this scenario the process of pulling data from production systems and populating marts or warehouses is complex. With one or two data stores, companies can simply hard-code the connection to the relevant databases and access data as required. Any changes to structures and formats can be communicated from the department and incorporated into the code with few problems. There are also tools that manage the complex task effectively.
In the real world, however, chances are data is going to be stored in multiple systems and in various locations. Hard-coding data connections becomes more difficult and it is less feasible to connect the data-hungry application directly to each source along with the necessary security considerations for each system. The alternative is a mammoth consolidation and integration project across the whole company. Apart from the time and cost involved (not to mention calculating the chances of success) in such a project, many companies prefer not to risk disrupting operations and to leave data in the hands of local owners.
Another, more modern data option is to create an application capable of pulling data from the relevant databases as it is called for. Such an application focuses on breaking down the data request into smaller queries according to the data stores it needs to access for data. A good idea in theory, but the process is so complicated it is viable only to companies with the financial and technical resources to manage this complexity.
Enter the data service layer
With companies now globalising, more executives are realising the importance of data and the necessity to make it available across departmental boundaries.
Julian Field, MD, Centerfield Software
Closely aligned to a service-oriented architecture, the data service layer is designed to provide data to applications and users as requested. The idea is to develop a layer of services that will hide the complexity involved in accessing multiple, heterogeneous data sources and cut down on the number of connectors and APIs that have to be used.
Irrespective of the location or format of the data, the data service delivers the information requested to the users via a uniform access mechanism. In other words, all the users need to learn is how to program requests to the data service layer, leaving it to go through the complexities of collecting and presenting the right information in the appropriate format.
The service also allows data owners in various silos to manage their information as they see fit and then register the data they have available. Other users can then access the information as needed without compromising its security. As more data is registered, the service layer becomes an information broker, allowing applications to use it as a resource for finding all intelligence available in the organisation.
Of course, this only works if the whole company buys into the project and all data is registered for anyone to access. If one has to be realistic, however, a project like this will never work unless someone in the organisation is driving it, ensuring people commit to it and educating users on the benefits they can expect from the data service layer. Reporting back to all influencers and decision-makers on its effectiveness and what it has brought the company in real terms is also a critical role for this person.
From this job description, it`s obvious that the traditional IT drivers will not be able to manage this role as their attention will be stretched across various responsibilities. What companies need therefore, is a data manager, somebody responsible for all the organisation`s information.
With a dedicated data manager in place evangelising the benefits of a data service layer and making it a technical reality (not the simplest of tasks), organisations can virtually forget about integration and consolidation from a data perspective. Instead, management, users and applications will have access to the data required, irrespective of where it is located and in which database and format it is stored. As it is drawn from operational systems, all users need do is ask for it and the data service will deliver.
Share