Once these and other questions are answered, then and only then should the various data management technologies be considered. It is clear that a plethora of systems exist to properly house, access and protect each set of data while optimizing the value of that data to the businesses.
Finally and most importantly, an infrastructure must exist that can facilitate the provisioning of each of the systems that a company uses to manage data. The business needs change moment to moment and the need to provision systems, while efficiently using the available physical resources is paramount. The infrastructure must be elastic, flexible, self-healing, and navigable enough to allow for the applications to perform their own functions on the greater set of data without disturbing or gridlocking other applications which may be equally important.
Data should flow through the infrastructure from app to app as a fluid flows through a well-designed and well-constructed set of piping. Each application will access the data that it individually needs to perform its individual business function. If the app needs to access large amounts of data to be analyzed with extreme alacrity then that app will be able to leverage the infrastructure for the appropriate resources to complete the task. If the app’s function is to absorb massive amounts of new data and complete transactions with “ultra-low” latency then those transactions will be completed in memory and if those transactions must be stored long-term they may later be later loaded into an underlying RDMBS or to unstructured files on a large file store using a global namespace. If those same transactions must be archived through perpetuity then the appropriate archiving functions will be utilized.
A main pillar of the Unified Data Strategy, described in part 1 of this article is that all business functions will be considered and addressed prior to the implementation of the features of the various data management technologies. Those data technologies will be integrated and the subsequent cross traffic will be facilitated by an infrastructure that optimizes the data flow. In fact many apps will be indistinguishable from the data they operate on.
The common denominator throughout all these ambitious technological innovations and objectives is virtualization. Through the magic of virtualization and virtualized infrastructures each app will be allocated the exact amount of resources necessary to complete their business functions. Also, each app can be provisioned and expired as necessary.
Most importantly, the data must be analyzed so that individual subsets of the greater dataset can be categorized and conjoined with the business functions so as to comprehensively understand exactly how an individual dataset must be managed. The functional demands of the apps and the intrinsic value of the data must be concurrently recognized. The data management technology must manage the data according to its innate value while concurrently satisfying the application’s directive to address the intent of the business function. Each data oriented technology defines the style of the data management while satisfying the applications demand for that data. Consequently, the dataset should not be defined until the business function is clear. To accomplish these goals, a UDS and the products which constitute the system must be flexible, elastic and dynamic. The UDS must be inclusive of all types of data and capable of defining new datasets when new requests are presented. The UDS will also maintain comprehensive yet distributed control over the tangible manifestation of the data in all forms.
Regardless of how the system works the primary evolutionary change in data management is being driven by a change in human and business culture. Data will no longer be solely the province of the massively expensive but functionally limited products that dominated the second half of the 20th century. Data will be managed, organized, monitored and protected according to business requirements and the actual objective value of any particular data set. It is at this point that the business will be in control of the data rather than the data being in control of business.
Part 3 of this article will concentrate on the science behind a UDS as well the mechanism for successfully operationalizing a UDS. The new science of Data Persona Analytics (DPA) and the 21st century notion of commoditized “attention” will be introduced into the IT lexicon.