Diversified Data Views
Data views are tools for BI and data professionals as well as end users. For professionals, the ultimate goal is to create and share flexible views for end users. For end users, flexible interactivity in professionally created views is a great starting point. But more importantly, the environment must allow them to grow beyond pre-designed views in order to achieve self -service. Data views, therefore, need multi-faceted capabilities to serve different stakeholders.
Iterative & Intertwined Professional Design
For professionals, data visualizing is a natural companion to data modeling. These two functions are intertwined and are used iteratively for experimenting and refining. This process not only produces powerful end user views such as dashboards but also generates reusable data models for applications beyond the originally conceived views. For example, a data model created for an executive dashboard can also be the base for generating a detailed, paginated report for operational staff.
Users' various responsibilities and skill levels dictate that they will need different levels of self-service. For casual users, a no-training-required, guided data view wizard could be the most productive starting point. For power users, they may want to expand upon existing views or create new views from existing data models. For data savvy users, they can introduce new data and mash it up with existing models. These self-service tools also need to be easily accessible, such as with one-click launch in a browser, so that users can dynamically experiment and grow into self-service "citizen developers" without much IT support.
By nature, data models are metadata which describe underlying source data. But performance can be a bottleneck if data is simply extracted from sources for data views. Self-service, ad hoc views can create other unanticipated operational performance issues. Therefore, a view must be materializable so that it actually caches data. However, a cache is not a simple duplication of data. It is optimized for the data views to enable rapid responsiveness and minimum resource consumption.