On June 15 2011, Forrester released a report(download the free report at Composite Software) for Enterprise Architects declaring that data virtualization is on the verge of delivering on the promise of information-as-a-service (IaaS). The report predicts the potential of substantial integration cost savings from successful adoption of the technology and provides three case studies to highlight the benefits. A data integration best practise recognized by Forrester is the adoption of what it calls an "hourglass architecture" in the implementation of the Information Fabric.
As per Forrester, one of the key requirements that will drive increased adoption is tactical demand for better data management and data integration alternatives to ETL or DBMS consolidation. Data virtualization abstracts, transforms, federates and delivers data from a variety of sources and presents itself as a single access point to a consumer regardless of the physical location or nature of the various data sources. How is this more beneficial than traditional ETL and DBMS approaches:
- flexibility and agility of integration due to the short cycle creation of virtual data stores without the need to touch underlying sources;
- improved data quality due to a reduction in physical copies; and
- improved usage through creation of subject-oriented, business-friendly data objects.
A key data integration design pattern revealed in the report as a result of the Forrester study is the "hourglass architecture". An hourglass architecture as defined by Forrester is a layered architecture with three key features: 1) A gradual shift from physical to virtual modalities as data moves closer to the users through the 2) use of intermediary canonical models and 3) a final virtual mapping layer to provide data to consumers in the expected format.
Inspite current data virtualization technology adoption at less than 20% of the IT organizations, Forrester predicts:
Over the next 18 to 36 months, we expect this market attitude to change as technology advancement, more third-party integration, and new usage patterns lead to increasing awareness of data virtualization’s potential. Already, many early adopters are having significant success with recent releases of the market-leading products. For example, one interviewee from Qualcomm stated: “When we realized that we didn’t have to physically move data around for integration, the technology started to really make sense. Now we have gone from point solutions to an enterprise deployment [of data virtualization].”
What is driving this change? Technological advances have addressed a number of the adoption concerns pertaining to difficult configuration, inability to meet performance requirements and security concerns:
Cost-based query optimization increases the number of usage patterns
Distributed caching enables enterprise-scale operations
Improved discovery tools make virtual data stores easier to create
Data masking adds element-level protection to virtual data sources
More out-of-the-box third-party integrations create a true enterprise platform
Integration of big data expands the potential for business insight
Does this instill confidence in you that the technology is ripe for adoption and precedes MDM, data governance or metadata management efforts? What are your concerns about this technology?